.. .flr Kin M‘ his arm...u.n¢:.'s I a... 5.. f: at}? .3 .1 . , 1. «Jam... o. , ‘ . if .35? imma‘mv.‘ . . :3”. vi. ‘ . ~ . .18. 44 v3.7 . .: y 4 . I. fin 3.5 in; kah x x. J .\ Jun... .32.? 4.35.5... a 59.3%.“me mm... A =va 7.. .fiwflfih 0 ~ i ; zawwua i . v Vfl. h. wiepud . . .1. . 1.9%.... .3 ”at“. is». n‘. 311x133?! 2.1.2:...fl Hananwnihmw » r . x. an: x .. .w 3‘ ~ 1:. ".15.. ‘1 .2 Ar: : .. .l .3 . 1t. nu zeta a... r....;._. s” .1‘ ._..».ru.Iw..‘ V72: 3 v. ._. «'52... 3.5. .4 .. . fi an? 3.93%. ,Xa.:¥....afiw§:§% L hammwdhufi§§ wk swuufié- mam LIBRARY 0? Michigan State University float» This is to certify that the dissertation entitled THE ECOLOGY AND EVOLUTION OF ELEVATION RANGE LIMITS IN MONKEYFLOWERS (MIMULUS CARDINAL/S AND M. LEWIS/I) presented by AMY LAUREN ANGERT has been accepted towards fulfillment of the requirements for the PhD. degree in Plant Biology km?” W . MM U Major Professor's Signature dg/Oé/of Date MSU is an Affirmative Action/Equal Opportunity Institution PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. MAY BE RECALLED with earlier due date if requested. DATE DUE DATE DUE DATE DUE 7937 A g e 3 0 If 5 2009 Mmm.mia It“ ' . lilll'“ llll W ii! l .., .4 ilr'lir it"s; f“. ‘ .. .“i "TT it. m "i i= .Mt‘nfluiu; ALI LL .1. THE ECOLOGY AND EVOLUTION OF ELEVATION RANGE LIMITS IN MONKEYFLOWERS (MIMUL US CARDINALIS AND M. LE WISII) By Amy Lauren Angert A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Plant Biology Program in Ecology, Evolutionary Biology, and Behavior 2005 ABSTRACT THE ECOLOGY AND EVOLUTON OF ELEVATION RANGE LIMITS IN MONKEYFLOWERS (MIMUL US CARDINALIS AND M. LE WISII) By Amy Lauren Angert Living organisms inhabit an incredible array of environments across the planet, but any particular species occurs in only a subset of habitats and geographic areas, an observation so fimdamental that its cause is rarely questioned. Nevertheless, the ecological and evolutionary forces that give rise to species’ distribution limits remain poorly understood. To determine whether species are maladapted to the environment at and beyond the distribution boundary, I investigate how fitness changes across the elevation ranges of closely related species of monkeyflower, Mimulus cardinalis and M. lewisii (Phrymaceae) in the Sierra Nevada Mountains, California. I use transition matrix models to estimate asymptotic population growth rates and find that population growth rates of M. Iewisii are highest at the range center and reduced at the range margin. Population growth rates of M. cardinalis are highest at the range margin and greatly reduced at the range center. Because observations of natural populations cannot determine fitness beyond a species’ present distribution, I reciprocally transplanted M cardinalis and M. lewisii within and beyond their present elevation ranges. For both species, I find the greatest average fitness at elevations central within the range, reduced fitness at the range margin, and zero or near-zero fitness when transplanted beyond the present elevation range limit. To identify the underlying causes for changes in fitness versus elevation, I examine plant performance in growth chambers simulating low and high elevation temperature regimes and show that temperature alone generates patterns of differential survival and growth similar to those observed in reciprocal transplant gardens. Mimulus Iewisii and M. cardinalis differ in photosynthetic physiology under temperature regimes characterizing their contrasting low and high elevation range centers, suggesting that the species’ elevation range limits may arise, in part, due to metabolic limitations on growth that ultimately decrease survival and limit reproduction. To measure natural selection on physiological and phenological traits within and beyond elevation range limits, I transplanted interspecific hybrids to low and high elevation and find that selection favors early flowering at high elevation and increased leaf photosynthetic capacity in warm temperatures at low elevation. I also find that hybrids selected at high elevation display reduced biomass when grown in temperatures characteristic of low elevation, suggesting that adaptation to the environment within the range may entail a cost to adaptation in other environments that places evolutionary constraints on range expansion. ACKNOWLEDGEMENTS I thank D. Schemske for his support and encouragement during all stages of my project. I also thank my committee members, H. Bradshaw, J. Conner, and K. Gross, as well as University of Washington committee members, R. Huey, J. Kingsolver, and J. Maron for their input. I am grateful to J. Anderson, K. Brady, B. Clifton, D. Ellair, D. Grossenbacher, A. MacMillian, and A. Wilkinson for assistance with data collection; P. Stone and family for use of their property and endless hospitality, M. Bricker, D. Ewing, and M. Hammond for plant care; S. Beatty, L. Ford, J. Haas, C. Millar, and P. Sterbentz for assistance obtaining research permits; C. Horvitz for Matlab code, and B. Igié, K. Kay, H. Maherali, J. Ramsey, and J. Sobel for many helpful discussions. Financial support was provided by a National Science Foundation Graduate Research Fellowship and Doctoral Dissertation Improvement Grant, a Sigma Xi Grant-In-Aid of Research, a California Native Plant Society Fellowship, a Northwest Orchid Society Fellowship, and the University of Washington Botany Department. iv TABLE OF CONTENTS LIST OF TABLES .............................................................................. vii LIST OF FIGURES ............................................................................ xiii CHAPTER 1 DEMOGRAPHY OF CENTRAL AND MARGINAL POPULATIONS OF MONKEYFLOWERS (MIMUL US CARDINALIS AND M. LE WISII) Abstract .................................................................................... 1 Materials and Methods .................................................................... 5 Results .................................................................................... 19 Discussion ................................................................................. 34 CHAPTER 2 VARIATION IN FITNESS WITHIN AND BEYOND MIMUL US CARDINALIS AND M. LE WISII ELEVATION RANGES Abstract ..................................................................................... 43 Materials and Methods ................................................................... 48 Results ...................................................................................... 59 Discussion ................................................................................... 67 CHAPTER 3 GROWTH AND LEAF PHYSIOLOGY OF MONKEYFLOWERS (MIM UL US CARDINALIS AND M. LE WISII) WITH DIFFERENT ELEVATION RANGES Abstract ..................................................................................... 77 Materials and Methods .................................................................. 81 Results ...................................................................................... 88 Discussion ................................................................................... 94 CHAPTER 4 NATURAL SELECTION WITHIN AND BEYOND THE ELEVATION RANGES OF MONKEYFLOWERS (MIM UL US CARDINALIS AND M. LE WISII) Abstract ................................................................................. 104 Materials and Methods ............................................................... 109 Results .................................................................................. 1 16 Discussion .............................................................................. 127 APPENDIX A .................................................................................. 134 APPENDIX B .................................................................................. 137 APPENDIX C .................................................................................. 141 LITERATURE CITED ........................................................................ 150 vi LIST OF TABLES Table 1. Generalized linear mixed models of the effects of range position, location within range position, and yearly transition interval on annual survival. Survival was modeled with a binomial distribution and a logit link function. F tests for fixed effects constructed by SAS GLIMMIX procedure. Z values for random effects obtained by dividing each variance estimate by its approximate standard error. No values remained significant after sequential Bonferroni adjustment to maintain table-wide type I error of 0.05 for each species (Rice 1989) ............ 21 Table 2. Log-linear analyses of the effects of location, L, and year, Y, on fate, F, for each stage class, S. Summing the effects of each stage class gives the four-way model of the effects of location and year on the state-by-fate transition table, conditional on the differences in states among locations and years. Notation follows convention for hierarchical models, such that the presence of an interaction implies the presence of all lower-order interactions and single factor terms contained in the interaction. Values in boldface remain significant after sequential Bonferroni correction to maintain a table-wide type I error rate Of 0.05 for each species (Rice 1989) ....................................................... 23 Table 3. Results of mixed model analysis of variance testing the effects of range position, year, and population within range position on reproduction. Random effects denoted by ‘[R]’. Flower number log-transformed and fruit set (proportion of flowers maturing fruit) arcsine square-root transformed prior to vii analysis. F -tests for fixed effects constructed by SAS MIXED procedure, with denominator degrees of freedom obtained from the Satterthwaite approximation and indicated in parentheses below each F -value. x2 values for random effects from likelihood ratio tests. Values in bold remain significant after sequential Bonferroni adjustment to maintain table-wide type I error of 0.05 for each species (Rice 1989) ....................................................................... 26 Table 4. Linear regressions of stage class density (2001-2003 mean number of plants per m2) and temporal variation in stage class density (coefficient of variation, CV, in 2001-2003 density) versus elevation along 50-200 m transects. After sequential Bonferroni correction to maintain a table-wide type I error rate of 0.05, no regression coefficients differed from zero ................................... 35 Table 5. Analysis of accelerated failure-time models for survival time, using 1339 uncensored values and 1273 right-censored values for M. cardinalis, 1339 uncensored values and 1073 right-censored values for M. lewisii, and a Weibull distribution .................................................................................. 58 Table 6. Pairwise differences of transplant site regression coefficients from accelerated failure-time analyses. Afier correcting for multiple comparisons, only Z-scores > 2.12 remain significant at the 0.05 level ................................................ 59 Table 7. Linear mixed model analysis of variance summary for log-transformed average annual stem length. F -tests for fixed effects constructed by SAS MIXED procedure, with denominator degrees of freedom obtained from the Satterthwaite approximation and indicated in parentheses below each F-value. viii All random effects (sire, dam, and their interactions) were estimated to be zero or near-zero and were not significant ................................................... 61 Table 8. Generalized linear mixed model analyses Of average annual fitness. F-tests for fixed effects constructed by SAS MIXED procedure, with denominator degrees of freedom obtained from the Satterthwaite approximation and indicated in parentheses below each F-value. Z-tests for random effects constructed by ‘covtest’ option ............................................................................ 64 Table 9. Rank correlation between population average annual fitness and I transplant elevation — population origin elevation I .............................................. 66 Table 10. July 2002 mean temperatures recorded in reciprocal transplant gardens at 415 and 2395 m ................................................................................ 84 Table l 1. Linear mixed model analysis of variance summary for four leaf physiological traits: instantaneous net photosynthetic rate (A), effective quantum yield ((Dpsu), stomatal conductance (gs), and the ratio of intercellular to ambient C02 (Ci/Ca) gs was corrected for temperature-induced changes in water viscosity and log- transforrned prior to analysis. F-tests for fixed effects constructed by SAS MIXED procedure, with denominator degrees of freedom obtained from the Satterthwaite approximation and indicated in parentheses below each F ~value. All random effects (population nested within elevation of origin, family nested within population, and their interactions with temperature) were estimated to be zero or near-zero and were not significant... Abbreviations as follows: Temp. = temperature, Spp. = species, Elev. = elevation ........................................ 89 ix Table 12. P-values from single degree of freedom independent contrasts of least square means testing the null hypotheses that interspecific differences in physiological parameters within a temperature regime and intraspecific differences between temperature regimes are equal to zero. gs was corrected for temperature-induced changes in water viscosity and log-transformed prior to analysis... . .. . .. . 8.9 Table 13. Linear mixed model analysis of variance summary for stem height and log- transformed aboveground biomass. F—tests for fixed effects constructed by SAS MIXED procedure, with denominator degrees of freedom obtained from the Satterthwaite approximation and indicated in parentheses below each F-value. All random effects (population nested within elevation of origin, family nested within population, and their interactions with temperature) were estimated to be zero or near-zero and were not significant. Symbols and abbreviations as in Table 11 .................................................................................... 93 Table 14. P-values from single degree of freedom independent contrasts of stem length and aboveground biomass least square means testing the null hypotheses that interspecific differences in grth parameters within a temperature regime and intraspecific differences between temperature regime are equal to zero .......... 94 Table 15. Survival and flowering of parental species and interspecific hybrids at low elevation (Jamestown, 415 m) and high elevation (White Wolf, 2395 m) sites in the central Sierra Nevada Mountains, California. Data recorded at Jamestown after one growing season and at White Wolf after three growing seasons. . ....1 17 Table 16. Logistic regressions of the probability of survival and probability of flowering for M. cardinalis, M. lewisii, and interspecific hybrids grown at low (Jamestown, 415 m) and high (White Wolf, 2395 m) elevation. All differences remain significant after sequential Bonferroni adjustment ........................ 117 Table 17. Correlation matrices of traits measured on hybrids grown at low and high elevation temperature regimes in growth chambers. Correlations within the high elevation temperature regime given above the diagonal. Correlations within the low elevation temperature regime given below the diagonal. Flowering was not observed in the high elevation temperature regime and freeze damage was not observed in the low elevation temperature regime (indicated by “n/a” = not applicable). Asterisks indicate significant Pearson correlation coefficients (TP < 0.10, *P < 0.05, "P < 0.01, ***P < 0.001, ****P < 0.0001). Sample size given in parentheses below each coefficient ................................................ 123 Table 18. Analysis of variance summary for physiological and phenological traits and fitness components of hybrids grown in a low elevation temperature regime. For each variable, the effect of selection regime (low elevation, high elevation, or greenhouse control) was tested with one-way ANOVA. Values in boldface remain significant after sequential Bonferroni correction (Rice 1989) .......... 124 Table 19. Analysis of variance summary for traits and fitness components of hybrids grown in a high elevation temperature regime. For each trait, the effect of selection regime (low elevation, high elevation, or greenhouse control) was tested with one-way ANOVA. Stem length was log-transfonned and percent tissue damage was arcsine square-root transformed prior to analysis. No values remain significant after sequential Bonferroni correction (Rice 1989). ..........124 xi Table A1. Species present, location (county, nearest landmark), watercourse, latitude and longitude coordinates (°N, °W), elevation (m), and plant density (individuals per m2) of populations chosen for detailed demographic study (D), transects of stage class abundance (T), reciprocal transplants (R), or incubator experiments (1). Abbreviations as follows: Species: c = M. cardinalis, l= M. lewisii; County: Ma = Mariposa, Tu = Tuolumne ...................................................... 135 Table C1. Transition matrices, sensitivity values, and A for each species, location, and yearly transition interval. Stage class abbreviations as follows: “Sm. nr.” = small non-reproductive (M. cardinalis: stem length 5 3 cm; M. lewisii: stem length 5 5 cm), “Lg. nr.” = large non-reproductive (M. cardinalis: stem length > 3 cm; M. lewisii: stem length > 5 cm), and “Repro.” = reproductive. Bias-corrected 95% confidence intervals for A (Caswell 2001) given in parentheses below each value ...................................................................................... 142 xii LIST OF FIGURES Figure 1. Spatial variation in proportion survival of each stage class. Data presented are means (over all yearly transition intervals) + SE. Stage class abbreviations as follows: Sm. non. = small non-reproductive, Lg. non. = large non-reproductive, Reprod. = reproductive. Location abbreviations as follows: C = central location, M = marginal location, BU = Buck Meadows (830 m), RP = Rainbow Pool (833 m), WA = Wawona (1208 m), CA = Carlon (1320 m), ML = May Lake (2690 m), WF = Warren Fork (2750 m) ........................................................ 20 Figure 2. Spatiotemporal variation in reproduction. a) M. cardinalis flower number, b) M. lewisii flower number, c) M. cardinalis fruit set, (1) M Iewisii fruit set, e) M cardinalis seed number per fi'uit, and f) M. lewisii seed number per fruit. Data presented are means + SE. Location abbreviations as in Figure 1. Year abbreviations as follows: 00 = 2000, 01 = 2001, 02 = 2002, 03 = 2003.... . . .....28 Figure 3. Asymptotic population growth rates (,1) for each location and transition interval and for pooled location matrices. Vertical bars indicate bias-corrected 95% confidence intervals (Caswell 2001). Asterisks indicate significant among- year variation within a location based on randomization tests (*P<0.05, **P<0.01, ***P<0.001, ****P<0.0001). Pooled A not sharing letters differ significantly from one another (after sequential Bonferroni adjustment) based on randomization tests. Abbreviations as in Figures 1 and 2 ........................... 29 xiii Figure 4. Life table response experiment (LTRE), with effects of location, yearly transition interval, and the interaction of location and year on 2. Effects obtained by summing the contribution of each transition matrix element to variation in lambda. Vertical bars indicate bias-corrected 95% confidence intervals (Caswell 2001). Abbreviations as in Figures 1 and 2 ......................................... ...31 Figure 5. Transition matrix element contributions to spatial variation in ,1. Vertical bars indicate bias-corrected 95% confidence intervals (Caswell 2001). Location abbreviations as in Figure 1. Stage class abbreviations as follows: sd = seed, sn = small non-reproductive, ln = large non-reproductive, re=reproductive ...... Figure 6. Schematic transect of the central Sierra Nevada Mountains, California, showing M Iewisii and M cardinalis elevation ranges and placement of reciprocal transplant gardens, after Clausen et a1. (1948) Figure 7. Survivorship at each transplant site. A) M cardinalis. B) M lewisii. ...33 50 Transplant site abbreviations as follows: JA = Jamestown, MA = Mather, WW = White Wolf, T1 = Timberline ............................................................ 60 Figure 8. Species average annual stem length + SE at each transplant site (mean values given within each bar). A) M. cardinalis B) M. lewisii. Site means sharing the same letter are not significantly different. Note that species are graphed on different scales. Transplant site abbreviations as in Figure 7 .................... Figure 9. Population average annual stem length + SE versus transplant site. A) M. cardinalis. B) M Iewisii. Populations are arrayed in order of increasing elevation of origin. Population means sharing the same letter are not significantly different. Note that species are graphed on different scales. xiv ...62 Transplant site abbreviations as in Figure 7. Population abbreviations as follows: Ma = Mariposa Ck., Mo = Moore Ck., Be = Bear Ck., Sn = Snow Ck., Te = Tenaya Ck., Tu = S. Fork Tuolumne R., Ta = Tamarack Ck., Po = Porcupine Ck., Ti = Tioga Rd., Sn = Snow Ck., Wa = Warren Fork Lee Vining R.. . . . .....62 Figure 10. Species average annual fitness (in units of flowers per year) + SE versus transplant site (mean values given within each bar). A) M cardinalis B) M Iewisii. Site means sharing the same letter are not significantly different. Note that species are graphed on different scales. Transplant site abbreviations as in Figure 7 ..................................................................................... 65 Figure 11. Population average annual fitness + SE versus transplant site. A) M cardinalis. B) M lewisii. Populations are arrayed in order of increasing elevation of origin. Population means sharing the same letter are not significantly different. Note that species are graphed on different scales. Abbreviations as in Figure 9 ................................................................................ 65 Figure 12. Comparisons of species mean + SE physiological responses to low elevation (hot) and high elevation (cold) temperature regimes: a) net photosynthetic rate (A, in umol co2 m‘2 s"), b) effective quantum yield [ops]. =(Fm. — Fs)/Fm'], c) stomatal conductance (gs, in mol H20 rn‘2 s", corrected for increased water viscosity at high temperature), and d) the ratio of intercellular to ambient C02 (Ci/Ca) ....................................................................................... 91 Figure 13. Species’ mean + SE a) stem length (cm) and b) aboveground biomass (g)...94 Figure 14. Relative fitness of parental species and hybrids transplanted to low (Jamestown, 415 m) and high (White Wolf, 2395 m) elevation .................. 118 XV Figure 15. Seed number per fruit versus pollination date for M lewisii and hybrids grown at high elevation (White Wolf, 2395 m). Regression coefficients (b) from linear regression analysis (*P < 0.05, ***P < 0.001) ............................... 119 Figure 16. Mean (+ SE) trait values of M cardinalis and M lewisii in the low elevation temperature regime. a) stem length, b) aboveground biomass, C) instantaneous net photosynthetic rate, d) stomatal conductance, e) effective quantum yield, and f) intercellular: ambient C02. Results of one-way ANOVA testing the effect of species given for each trait. All values remain significant after sequential Bonferroni adjustment to maintain a type I error rate of 0.05 ..................... 120 Figure 17. Mean (+ SE) trait values of M cardinalis and M lewisii in the high elevation temperature regime. a) stem length, b) aboveground biomass, c) instantaneous net photosynthetic rate, d) stomatal conductance, e) effective quantum yield, and f) intercellular: ambient C02. Results of one-way ANOVA testing the effect of species given for each trait. The difference in biomass remains significant after sequential Bonferroni adjustment to maintain a type I error rate of 0.05.. .......121 Figure 18. The effect of selection regime (low elevation, greenhouse control, or high elevation) on two leaf physiological traits and flowering phenology of hybrids grown in a temperature regime characteristic of low elevation. Values given are mean + SE a) effective quantum yield, b) stomatal conductance, and c) days to first flower. P-values in boldface remain significant after sequential Bonferroni correction. Hybrid means not sharing letters differ significantly based on Tukey- Kramer adjusted comparison of least square means ................................ 125 xvi Figure 19. The effect of selection regime (low elevation, greenhouse control, or high elevation) on fitness components of hybrids grown in a temperature regime characteristic of low elevation. Values given are mean + SE a) aboveground biomass, b) stem length, 0) proportion survival, and d) flower number. P-values in boldface remain significant after sequential Bonferroni correction. Hybrid means not sharing letters differ significantly based on Tukey-Kramer adjusted comparison of least square means ..................................................... 126 Figure 20. The effect of selection regime (low elevation, greenhouse control, or high elevation) on two leaf physiological traits and post-freeze tissue damage of hybrids grown in a temperature regime characteristic of high elevation. Values given are mean + SE a) effective quantum yield, b) stomatal conductance, c) % tissue damage following freeze 1, and d) % tissue damage following freeze 2. The effect of selection regime on tissue damage following freeze 1 does not remain significant afier sequential Bonferroni correction .......................... 128 Figure 21. The effect of selection regime (low elevation, greenhouse control, or high elevation) on fitness components of hybrids grown in a temperature regime characteristic of high elevation. Values given are mean + SE a) aboveground biomass, b) stem length, and c) proportion survival ................................ 129 xvii CHAPTER 1 Demography of central and marginal populations of monkeyflowers (Mimulus cardinalis and M. lewisii) Abstract—Every species occupies a limited geographic area, but how Spatiotemporal environmental variation affects individual and population fitness to create range limits is not well understood. Because range boundaries arise where, on average, populations are more likely to go extinct than to persist, range limits are an inherently population- level problem that require a demographic framework. In this study, I compare demographic parameters and population dynamics between central and marginal populations of monkeyflowers, Mimulus cardinalis and M lewisii, along an elevation gradient spanning both species’ ranges. Central and marginal populations of both species differed marginally in survival and significantly in fecundity. For M Iewisii, these components of fitness were higher in central than in marginal populations, but for M cardinalis the converse was true. To assess Spatiotemporal variation in population dynamics, I used transition matrix models to estimate asymptotic population growth rates (A) and found that population growth rates of M lewisii were highest at the range center and reduced at the range margin. Population growth rates of M cardinalis were highest at the range margin and greatly reduced at the range center. During the study period, temporal variation in A was of smaller magnitude than spatial variation in 2.. Using life table response analysis, I decomposed Spatiotemporal variation in ,1 into contributions from each transition between life stages and found that transitions from large non-reproductive and reproductive plants to the seed class and stasis in the reproductive class made the largest contributions to spatial differences in A. These transitions had only low to moderate sensitivities, and sensitivity values were largely similar across all locations, indicating that differences in projected population growth rates resulted mainly from observed differences in transition matrix parameters and their underlying vital rates. Continued study of spatiotemporal variation in population dynamics, in combination with estimates of dispersal between central and marginal populations, will improve our understanding of the species’ distribution limits. Key words: population dynamics, range limit, matrix population models, life table response experiment Every species occupies a limited geographic area. Sometimes ranges end at obvious environmental discontinuities, but more often ranges end at “seemingly arbitrary” points along gradual environmental gradients (Kirkpatrick and Barton 1997). Linking spatial and temporal variation in the environment to variation in both individual and population fitness is critical to understanding species’ distribution limits (Holt and Keitt 2005). Because range boundaries arise where, on average, the probability of population extinction exceeds the probability of persistence, range limits are an inherently population-level problem for which a demographic framework is informative. Range margins are often assumed to be coincident with ecological margins, such that species reach the limit of their environmental tolerance at a range boundary and are maladapted to conditions beyond the range (Antonovics 1976; Lesica and Allendorf 1995). Consistent with this characterization, species abundance (i.e., local population density) ofien decreases with distance from the range center, presumably in response to an increasingly unfavorable environment (McClure and Price 1976; Svensson 1992; Telleria and Santos 1993; Brown et a1. 1996; but see Sagarin and Gaines 2002 and references therein). Observations of individual performance across the range frequently find lower survival of certain life history stages or reduced fecundity at the range margin relative to the range center (Marshall 1968; Pigott and Huntley 1981; McKee and Richards 1996; Garcia et a1. 2000; Jump and Woodward 2003). However, whether reductions in some fitness components impact population growth and persistence is not always evident. In some instances, reductions in individual performance alone, without consideration of its secondary effects on population dynamics, may be insufficient to explain the position of a range boundary (Prince and Carter 1985). Carter and Prince (1988) suggested that the small reductions in fecundity observed across the range boundary of Lactuca serriola are insufficient to explain failure to occur beyond its present distribution. If populations are not seed limited, reductions in fecundity may not translate into reduced recruitment and population growth rates (Turnbull et a1. 2000; Maron and Simms 2001). A decrease in one fitness component from the center to the edge of the range also may be mitigated by other differences. For example, a loss of migratory behavior may counterbalance lower juvenile fitness in southern marginal populations of the Iberian robin, Erithacus rubecula (Perez-Tris et al. 2000). In the aquatic plant Decodon verticillatus, vegetative reproduction may offset reductions in sexual reproduction in northern peripheral populations (Dorken and Eckert 2001), and conditional seed dormancy may ensure persistence of peripheral populations of wild barley, Hordeum spontaneum (Volis et a1. 2004). To thoroughly understand geographic range limits, components of performance must be integrated into models of population growth across species’ distributions (Pulliam 2000). Temporal patterns of variation, as well as the interaction between spatial position and temporal dynamics, are also important to understanding the dynamics of populations across species’ ranges (Ives and Klopfer 1997). Central populations might exhibit greater inter-annual variability if intrinsic rates of increase are high in optimal habitat or if regulation by biotic factors such as predation acts more strongly when population density is high (Williams et a1. 2003). Alternatively, marginal populations may be at or near the limit of environmental tolerance, and consequently more vulnerable to environmental fluctuations that exceed mean tolerance levels in some years, causing marginal populations to vary in size or age structure more than central populations (Gaston 1990; Brown et a1. 1995). Occasional climatic extremes have been observed to cause pulses of mortality or reproductive failure at northern range limits in populations of tropical trees (Olmsted et a1. 1993), pool frogs (Sjogren 1991), pied flycatchers (Jarvinen and Vaisanen 1984), and North American birds (Mehlman 1997). Williams et a1. (2003) found that marginal populations of three bird species were both less dense and experienced greater variability in density-independent population growth rates. If marginal populations are small and exhibit high variability, then they may be vulnerable to extinction (Curnutt et a1. 1996; Nantel and Gagnon 1999; Maurer and Taper 2002; Vucetich and Waite 2003). Several studies hypothesize that spatial gradients in extinction risk, colonization rates, and/or habitat availability can create stable range boundaries (Carter and Prince 1981; Lennon et a1. 1997; Holt and Keitt 2000; Maurer and Taper 2002). Other models suggest that marginal populations may be demographic sinks, sustained only by immigration from source populations at the range center (Pulliam 1988; Kawecki 1995; Curnutt et a1. 1996; Kirkpatrick and Barton 1997; Guo et a1. 2005; Holt et a1. 2005). Although time series and survey data have been used to examine spatiotemporal population variation across the range over long time periods and at broad spatial scales (Curnutt et a1. 1996; Mehlman 1997; Doherty et al. 2003; Williams et al. 2003), very few investigations have compared the demography of central versus marginal populations (but see Nantel and Gagnon 1999; Stokes et al. 2004; Volis et al. 2004). This paper presents a comparison of demographic parameters and population dynamics between central and marginal populations of sister species of monkeyflowers, Mimulus cardinalis and M Iewisii, along an elevation gradient spanning both species’ ranges. To assess spatiotemporal variation in population dynamics, I used several analyses based on transition matrix models. I estimated the asymptotic population growth rate (A) in central and marginal populations over three yearly transition intervals, and examined the sensitivity of A to perturbations in matrix elements. I used life table response experiments (Caswell 2001) to decompose spatiotemporal variation in A into contributions from each transition between life cycle stages. Specifically, this study investigated the following questions: 1) How do vital rates and population growth rates vary between central and marginal populations? and 2) Which life cycle transitions are responsible for observed differences in population growth rate between populations? MATERIALS AND METHODS Study System Mimulus cardinalis and M lewisii (Phrymaceae) are rhizomatous perennial herbs that grow along seeps and stream banks in western North America. Both species are self-compatible and animal pollinated (Hiesey et a1. 1971; Schemske and Bradshaw 1999). The species occupy different latitudinal and altitudinal ranges. Mimulus cardinalis occurs from southern Oregon to northern Baja California and from coastal California to Arizona and Nevada. Mimulus Iewisii is composed of two races, a northern form occurring from southern coastal Alaska to southern Oregon and eastward to the Rocky Mountains and a southern form occurring primarily in the Sierra Nevada Mountains of California (Hiesey et a]. 1971; Hickman 1993; Beardsley et al. 2003). Here I study only the Sierran form of M Iewisii. In California, M cardinalis occurs from sea level to 2400 m and M Iewisii occurs from 1200 m to 3100 m (Hickman 1993). In the Yosemite National Park region where this research was conducted, the species co-occur on larger watercourses between 1200 and 1500 m elevation (A. Angert, unpub. data). Repeated attempts to locate extant populations in the Yosemite region at the upper limits of the published Californian distributions were unsuccessful, so I consider 1200-1500 m to be the shared mid elevation range limit of both species. Populations of each species were monitored along an elevation transect from 373 m to 2750 m within 37.464 and 38.098 ° N latitude in Yosemite National Park and the surrounding Stanislaus, Inyo and Sierra National Forests (Appendix A). Although this transect represents only a small fraction of each species’ geographic range, it provides a gradient from elevation range center to elevation range margin for both species at a tractable scale. The Yosemite region of the Sierra Nevada Mountains offers a large area of undeveloped habitat in which to study species’ natural distributions. Six sites were selected for detailed demographic study based on elevation and habitat quality (Appendix A). Two sites were located at middle elevation (Wawona, 1208 m, and Carlon, 1320 m) where both species occur sympatrically at their range margin. The remaining four sites were located at the low (Buck Meadows, 830 m, and Rainbow Pool, 833 m) and high (May Lake, 2690 m, and Warren Fork, 2750 m) elevation range centers for M cardinalis and M Iewisii, respectively. Additional sites were selected across a continuous range of elevations for estimates of plant density along 50-200 m transects (Appendix A). Census Plots During July-August 2000, multiple census plots were established within each of the six demographic sites. Plots varied in size and number across sites due to differences in habitat and plant density (average plot size 103.4 m2, range 8 — 459 m2; average total plot area per site 800.7 m2, range 437.8 — 1160.5 m2). The plots were chosen to span natural environmental variation present at each site and to encompass areas suitable for all life history transitions so that together they are representative of performance at the site as a whole. The comers of each plot were marked with rebar to facilitate relocation in subsequent years and to establish an (x, y)-coordinate system for mapping plant locations. Within each plot every M cardinalis and/or M lewisii individual was mapped to the nearest 5 cm on the (x, y)-coordinate grid and marked with a unique number using an aluminum write-on tag wrapped around a nail except when rocks prevented tag placement. When tag placement was impossible, (x, y)- coordinates were used to identify individuals. Individuals were defined as discrete clusters of stems separated from other stems by at least ten cm except when stems had evidence of physical connection or were known to have arisen from multiple seedlings. Because both species are capable of clonal growth via rhizomes, a small number of stems marked as individuals, particularly at the beginning of the study, may have been ramets of the same genet. Following plot establishment in 2000, censuses were conducted twice per year in early summer and autumn from 2001 - 2003. The date of censuses varied each year with the timing of snowmelt and spring floods. The early summer census captured over- winter survival of plants recorded in the previous year and spring seedling germination. The autumn census captured survival, grth and reproduction during the growing season. At the early summer census, new seedlings were mapped on the (x, y)-grid and given an impermanent colored marker. At the autumn census, permanent aluminum tags were given to all surviving recruits from the early summer census and to any additional recruits not present at the summer census. Over the four years of this study (2000 — 2003), the fates of a total of 16,849 plants were recorded (Buck Meadows, 569; Rainbow Pool, 2,157; Wawona, 128 M Iewisii, 4,557 M cardinalis; Carlon, 1,537 M lewisii, 3721 M cardinalis; May Lake, 1,513; and Warren Fork, 2,667). Stem number, stern length, flowering status, and flower and fruit number of all plants was recorded each autumn. For each plant, up to 20 non-flowering and 20 flowering stems were measured from the ground to the base of the last pair of expanded leaves; all remaining stems were tallied and used to estimate total stem length based on the average stem length of the 40 measured non-flowering and flowering stems. Plant fecundity was estimated by multiplying the number of mature fruits per flowering plant by the population mean seed number per fruit. Each fruit contains approximately 500-2500 tiny seeds and flowering individuals may have hundreds of fi'uits. Each fall, two fruits were harvested from each of 10 individuals growing at least several hundred meters downstream of the census plots. Sampling downstream ensured that patterns of seedling emergence within the plots were not altered by seed removal. In the lab, samples of approximately 200 seeds per fruit were counted under a dissecting microscope and weighed to determine the relationship between seed mass and seed number. Seed number per fruit was then estimated from the total seed mass. Seed samples could not be obtained for M Iewisii at Carlon in 2002 or for M cardinalis at Rainbow Pool in 2002 and Buck Meadows in 2003, so average seed number per fruit across all other years at the particular location was used to estimate fecundity. Seed Dormancy Mesh pouches. — Variation in population size and persistence may be influenced by recruitment from a seed bank (V olis et al. 2004). Air-dried seeds of M cardinalis and M Iewisii remain viable for many years when stored at room temperature (pers. obs.). This suggests that a seed bank could play a role in seedling recruitment and population persistence. To ascertain whether significant seed dormancy exists in nature, a seed viability experiment was initiated at one central and one marginal site for each species in September 2001 (central: M cardinalis, Rainbow Pool; M lewisii, May Lake; marginal: both species, Carlon). Field-collected seeds were enclosed in 5 x 10 cm pouches made from fine mesh (“No Thrips”, 150 x 150 u opening size, Green-Tek, Inc., Edgerton, WI, USA), allowing the seeds exposure to air, water and light while preventing seed entry into or escape from the pouch. Approximately five hundred seeds were placed in each of four pouches per site. Each pouch was staked to the ground in the vicinity of a reproductive plant to ensure that experimental seeds experienced environmental conditions similar to naturally dispersed seeds. No germination was observed while seeds were in the pouch. Two pouches were removed from each site in autumn 2002 and 2003, with the exception of the central M cardinalis site where all pouches were destroyed by vandalism. Pouches were taken to Michigan State University where the contents were sieved to separate seeds from silt. Seeds were then placed on moistened soil and allowed to germinate in the greenhouse. After one year, germination was 4.9% (out of the initial 500) in M Iewisii seeds from May Lake, 15.2% in M Iewisii seeds from Carlon, and 7.8% in M cardinalis seeds from Carlon. Afier two years, germination was 8.6% in M Iewisii seeds from May Lake, 3.6% in M Iewisii seeds from Carlon, and 0.2% in M cardinalis seeds from Carlon, demonstrating that seeds of both species may remain dormant and viable for at least one year in the seed bank. PVC stations. — To obtain parameter estimates for seed survival, detailed studies of seed dormancy were initiated at one central and one marginal site per species in September 2002 (central: M cardinalis, Buck Meadows; M Iewisii, May Lake; marginal: both species, Carlon). At May Lake and Carlon, eight replicate 20 x 20 cm plots per species were excavated to a depth of 15 cm to remove any previously existing seed bank. In each plot, seed-free soil from above the floodplain was used to refill the excavated area. Four rings cut from sections of poly(vinyl chloride) pipe (10 cm diameter, 8 cm tall) were buried in each plot, leaving 1 cm of ring above ground level. Two rings per station were randomly assigned the seed treatment, in which approximately 2160 seeds by volume were added in September 2002, and the remaining 10 two rings served as “no-seed” controls. At Buck Meadows the design was identical except a seed shortage allowed only 1 seed treatment per station containing approximately 1250 seeds by volume. Each year seedlings were counted and removed with forceps from these rings, once in early summer after most emergence had occurred and again in the autumn to capture any additional germination. Numerous stations were lost to flooding, tree fall, and animal disturbance, including all stations at Carlon. Due to small remaining sample sizes, data were pooled within each site for species-specific estimates of seed survival parameters (Appendix B). Stage Classification Each population was classified into four stages present at the autumn census using biological criteria based on relationships between size, survival, and reproduction and examination of frequency distributions of stem lengths for different aged cohorts of plants. To facilitate comparisons among sites and years, classification criteria were developed using pooled data from all sites and years for each species. The boundary between small and large non-reproductive plants was defined as the midpoint between the median total stem length of first-year non-reproductives (i.e., seedlings) and the median total stem length of non-reproductives aged two and older (midpoint: 3 cm for M cardinalis, 5 cm for M Iewisii). A seedling class based on age alone was not retained because rapid first-year grth frequently caused first-year plants to surpass older plants in size. Only one reproductive stage class was used because differences in the size distribution of reproductive plants between sites created small sample sizes in some reproductive classes when reproductive plants were subdivided by size. Also, 11 survival of reproductive plants within each site was not related to total stem length, indicating that subdivision of the reproductive class was not warranted. Variation in Fates of Vegetative Plants To examine variation in annual survival among locations and years, I modeled survival of each stage class as a function of position within the range (center or margin), population nested within range position, yearly transition interval (2000-2001, 2001- 2002, or 2002-2003), and all interactions using a binomial distribution and a logit link function (PROC GLIMMIX, SAS, version 9, SAS Institute, Cary, NC, USA). Range position and year were considered as fixed effects and population within range position was considered as a random effect. To evaluate the significance of fixed effects, I used Type III estimable functions. To evaluate the significance of random effects, I tested whether the Z-value of each effect (its variance parameter divided by its approximate standard error) was different from zero (Juenger and Bergelson 2000). I report significance values for this and all other analyses with and without sequential Bonferroni adjustment to maintain a table-wide type I error rate of 0.05 for each species (Rice 1989, Moran 2003) To examine variation in transition probabilities among locations and years, I performed log-linear analyses (Horvitz and Schemske 1995; Caswe112001). All vegetative plants were classified into stage classes each year, including an extra class for dead plants. For each species, these analyses considered the following categorical variables: state (stage at time t: small non-reproductive, large non-reproductive, or reproductive), year (transition interval: 2000-2001, 2001-2002, or 2002-2003), location (M cardinalis: Buck Meadows, Rainbow Pool, Wawona, or Carlon; M Iewisii: Wawona, Carlon, May Lake, or Warren Fork), and fate (stage at time t+1: small non- 12 reproductive, large non-reproductive, reproductive, or dead). The first set of analyses examined each state separately to ask whether the fate of a particular state varied among years and locations using three-way contingency tables defined by year, location, and fate for each initial state and the null hypothesis that fate was independent of year and location, given the predetermined distribution of plants into year and location categories (Horvitz and Schemske 1995). The second set of analyses examined whether the entire state by fate transition table varied between locations and years using the four-way contingency table defined by state, year, location, and fate and the null hypothesis that fate was affected by state but was independent of year and location (Caswell 2001). Likelihood ratio tests, obtained from the difference in goodness-of-fit G2 values between two models that differed only in the factor being tested, were used to evaluate the significance of particular factors (Caswell 2001). Log-likelihood statistics for all three-way analyses were obtained with PROC CATMOD (SAS, version 8, SAS Institute, Cary, NC), using the LOGLIN option and adding 0.5 to all cell counts to avoid estimation problems caused by zeros (Horvitz and Schemske 1995). Log-likelihood statistics for four-way analyses were obtained by summing stage-specific G2 values from three-way analyses across all stages (Horvitz and Schemske 1995). Variation in Reproduction To examine spatiotemporal variation in reproduction, I analyzed the effects of range position (center or margin), population nested within range position, and year on the reproductive variables total flower number (of flowering plants only), fruit set (the proportion of flowers maturing seeds), and seed number per fruit using PROC MIXED (SAS, version 8, SAS Institute, Cary, NC). Range position and year were considered as 13 fixed effects and population within range position was considered as a random effect. Missing seed counts per fruit at some locations in some years prevented the analysis of interactions with year for the dependent variable seed number per fruit. To meet the assumptions of traditional linear analysis, flower number was log-transformed and fruit set was arcsine square-root transformed. Analyses of transformed data produced qualitatively similar results to generalized linear analyses using Poisson (flower number) or binomial (fruit set) distributions; I present traditional linear analyses for simplicity. To evaluate the significance of fixed effects, I used Type III estimable functions, which tolerate unbalanced samples, with denominator degrees of freedom obtained by Satterthwaite’s approximation. Likelihood-ratio tests, comparing each reduced model to the full model including all effects, were used to evaluate the significance of random effects. Construction of Matrix Models Transition matrix models of population dynamics were constructed using estimates of reproduction, seed dormancy, recruitment, and transition probabilities among the three vegetative stages. These calculations were performed for each location and transition year to generate a set of 12 location-year matrices per species. The calculations were also performed on data pooled across all years within each location to generate a set of 4 pooled location matrices per species. Due to small sample size (N=5) of large non-reproductive M Iewisii at Wawona during 2002-2003, estimates of transitions from the large non-reproductive stage class were obtained from average transition frequencies across all years at Wawona (as in Menges and Dolan 1998). 14 The projection matrix model for these analyses was a linear, time-invariant model of the form n(t+1) = A - n(t), where n(t) is a vector of stage-classified individuals in the population at time t, n(t+1) is the stage-classified vector of individuals at one time step in the future, and A is a 4 x 4 projection matrix of transition probabilities and stage-specific fecundities that shows how individuals in stage j at time t contribute to stage i at time t+1. The top left-hand corner, a1 1, is seed dormancy; other cells in the top row, a12 — a”, are fecundities (mean number of seeds produced by a reproductive plant at time t+1) weighted by the probability of an individual in class j at time t becoming reproductive at time t+l. Non- reproductive stages have a non-zero contribution to the seed class if they may become reproductive within one time step. Occasionally, rapid growth of spring germinants enabled them to reach the reproductive class by the autumn census, in which case the top lefi-hand comer is both seed dormancy and the seed contribution of newly germinated reproductive plants. Matrix Analysis The dominant eigenvalue of a projection matrix is the asymptotic population growth rate, A (Caswell 2001). Although other interesting demographic parameters such as the stable stage distribution or reproductive values may be obtained from matrix projection analysis, I chose to focus on ,1 as a synthetic measure of demographic success in each environment. 15 A fixed-design life table response experiment (LTRE; Caswell 2001) was used to model A of each species as a linear function of location, I, yearly transition interval, y, and their interaction, 1y: 1W): 100+ a") +130) + (afi)"y’ where a“) is the effect of the 1‘'1 level of the location treatment, fl“) is the effect of the yth level of the year treatment, and (away) is the interaction of the Ith location and yth year, measured relative to the projected grth rate of the reference matrix (--). The reference matrix can be obtained from an unmanipulated control or by combining data from all treatments into a mean (calculated by averaging transition frequencies) or pooled (calculated from pooled raw data) matrix (Miriti et a1. 2001). I chose to use a pooled reference matrix, which weighted observed transitions by their frequency in the entire dataset (Horvitz and Schemske 1995) and better approximated observed lambdas than a mean reference matrix. Treatment effects were estimated as ynzyn_yo 2 may.” — ag(°'))-(6l/6a,-j) | (Am + An flu) =11”) _ A") z Hag-(y) — ay("))°(6,i/6a,j) I (A('y)+ Am)” (wt/y) = Atty) . ’20-) _ at!) _ 130) ~ I .. .. ~ :(aU‘ y) — a,‘ ))°(6)./6a,j) | (AUy) + A( >y2 - a“) - W )/2 where the sensitivity of A to changes in a matrix entry, til/dag], was evaluated midway between the treatment and the reference matrices and obtained from the equation away- : vin-/, where v and w are the right and left eigenvectors of the matrix (Caswell 2001). 16 The above equations can be interpreted to mean that the effect of the treatments on population growth depends on both observed variation in matrix elements and the sensitivity of population growth to variation in those elements. The contribution of a particular matrix element at, to variation in ,1 may be low if a2, did not vary between treatments and/or if A is insensitive to variation in a”. A matrix element with high sensitivity may not contribute to variation in ,3. if the transition was unaltered by the treatments. Conversely, a matrix element with slight variation but high sensitivity may make a large contribution to variation in A. To assess uncertainty in population projections, I used bootstrapping to calculate bias-corrected 95% percentile confidence intervals around estimates of A, sensitivities, and LTRE contributions (Caswell 2001). Bootstrap calculations were designed to mimic the data structure used to generate matrix parameters. For example, individuals were stored as columns in a data array, where rows represented fates and fruit numbers, and randomly selected with replacement to generate a bootstrapped dataset of size equal to the population sample size. For each bootstrapped dataset, transition probabilities among vegetative classes, total fruit number at times t and H , and fruit number per reproductive at time t+1 were calculated. For estimates involving seed number, a seed number per fruit was drawn at random from the empirical cumulative probability distribution of seed number for each time (t-l , t, and t+1) and then multiplied by fruit number within every bootstrap replicate. The empirical probability distribution for seed number was derived from the average seed count per fruit from ten individuals per species, location and year. An estimate of seed dormancy was drawn at random from the cumulative probability distribution of seed dormancy, 17 which was assumed to be normally distributed with mean and standard deviation derived from estimates across multiple seed stations for each species. Non-parametric randomization tests based on random permutations of individuals between groups were used to test specific hypotheses about differences in ,1 among yearly transition intervals and between locations (Caswell 2001). To assess whether ,1 varied among yearly transition intervals within a location, individuals were randomly permuted among pairs of years, keeping sample sizes for each transition interval fixed (Fréville et a1. 2004). Transition frequencies and fruit counts for each permuted dataset were calculated as described for each bootstrapped dataset above. Mean seed counts at times t-1, t, and t+1 were permuted independently, then combined with transition frequencies and fruit counts to generate matrices and calculate A for each transition interval for each of 2000 permuted datasets. The significance of the observed standard deviation of 11 among yearly transition intervals within each location was then compared to the randomized distribution of standard deviations with a one-tailed test (Caswell 2001; F réville et al. 2004). To assess whether 2 differed between locations, data were first pooled over all years within each location. Individuals, with their complete histories, and mean seed counts were randomly permuted between locations in a similar fashion to permutations among transition intervals as described above. The observed absolute value of the difference in ,1 was compared to the randomized distribution of absolute differences with a two-tailed test (Brys et a1. 2004). Because six pairwise comparisons of locations were made for each species, significance levels were adjusted according to the sequential Bonferroni procedure (Rice 1989; Edgington 18 1995). All matrix calculations were performed in Matlab, version 6.1 (The MathWorks, Natick, MA, USA). Transects Fourteen (M cardinalis) and sixteen (M Iewisii) additional census transects per species were established across a continuous range of elevations to examine spatiotemporal variation in local population density and to ensure that inferences about variation in population dynamics across the elevation ranges of M cardinalis and M Iewisii were drawn from a representative sample of central and marginal sites (Appendix A). Along each 50 — 200 m transect, every small non-reproductive, large non-reproductive, and reproductive individual of each species was tallied in autumn 2001, 2002 and 2003. The area of suitable habitat along each transect was estimated to correct for variation across sites in habitat availability. Density of each stage class was expressed as the number of individuals per m2. Linear regression models were used to examine variation in mean (over 2001-2003) stage class density versus elevation (PROC REG, SAS, version 8, SAS Institute, Cary, NC). To examine temporal variation in local population density, coefficients of variation in stage class density across years were calculated and regressed against elevation. The sequential Bonferroni procedure was used to maintain a table-wide type I error rate of a = 0.05 for each species (Rice 1989) RESULTS Spatiotemporal Variation in Fates of Vegetative Plants Small non-reproductive plants showed the lowest annual survival (M cardinalis: 11 — 22%, M Iewisii: 7 — 26%), and reproductive plants showed the highest annual 19 A) M cardinalis B) M Iewisii To .E1.0'— C,BU 1.0‘_C,m 30%: an 084— C,WF 3' —M,WA ' —M,WA 3 . €06“: M,CA 0.6“: M,CA § 0.4 . 0.4 . E 9. 0.2 ~ 0.2 ~ 9.. 0.0 ‘ 0.0 * Sm. nr. Lg. nr. Reprod. Sm. nr. Lg. nr. Reprod. Stage class Stage class Figure 1. Spatial variation in proportion survival of each stage class. Data presented are means (over all yearly transition intervals) + SE. Stage class abbreviations as follows: Sm. non. = small non-reproductive, Lg. non. = large non-reproductive, Reprod. = reproductive. Location abbreviations as follows: C = central location, M = marginal location, BU = Buck Meadows (830 m), RP = Rainbow Pool (833 m), WA = Wawona (1208 m), CA = Carlon (1320 m), ML = May Lake (2690 m), WF = Warren Fork (2750 m). survival (M cardinalis: 72 — 91%, M Iewisii: 81 - 97%; Figure 1). Position within the elevation range affected survival of reproductive plants of both species and marginally affected survival of M Iewisii large non-reproductive plants, although these effects did not remain significant after sequential Bonferroni correction (Table 1). Survival of M cardinalis reproductive plants was higher at the range margin than at the range center, whereas survival of M Iewisii reproductive plants was higher at the range center than at the range margin (Figure 1). Year did not affect annual survival, and the interaction of year and range position affected M Iewisii small non-reproductive plants only. The random effects of population within range position and the interaction of population and year were not related to annual survival of any stage class (Table 1). Log-linear analyses of transition probabilities for each stage class revealed that, for both species, year significantly affected the fate of non-reproductive but not 20 28.0 8.0 e 8.0 88.0 3.0 0 Rd 0 32:85 0%...8083 0 500.0 Nd H No.0 00NNd 0N4 a 00.0 300003 000003 000805 m 000800m m 000.00%.— 8005 m 33.0 00.0 0506 No.0 3 Ed 00.: N 00>...00Emom 00006 E A amnvd 36 302 .o $.m N 00> 003.0 3.2 mucod N02 00000 23 _ cosmmom m m m m m m 00% .0333 .3 802. N00 0" cod 33.0 2.0 H 3.0 mmmfio $0 0 and €0Emonc H00%...0000004 $00.0 mod .1." 8.0 0 0000.0 mmd 0 mod 300003 0000004 0 000.00%. 0 0083mm m 000800m 8003 $3.0 00.0 12.5.0 NNd 003.0 N _ A N 00»...:0060& Name woN good one N300 mwN N 00> $00.0 oN.oN NNm_.o 3.0 33.0 cod _ 00503 m m m m m m .0 005m 302.0000 .2 0300000000000: 0300000500000 0>00000a0m 0wu0q =08m 800.005 0.060% .Aawfl 003v 006000 0000 00.“ 30 m0 00000 H 093 0005-038 500508 00 000803.80 E0b0mcom 00000000 0000 Hgomefi 0050800 m020> 02 00000 00000000 000808500 0: 3 0008000 00:00? 0000 $0020 3 005050 $00b0 800000 00.0 m030> N 03000000 032230 mm>.8m ._0>_>.Sm 03:00 :0 0020:: 003300. >100» 0:0 £09000 0w=00 3505 000002 50000.00 0mg 00 8.00%? 05 .00 20008 0058 0000: 0002000000 ._ 030g. 21 reproductive stages and location significantly affected the fate of all three stage classes (Table 2). Marginal and conditional tests of the effects of location and year produced very similar results. Log-linear analyses of the four-way contingency table of state by fate transitions across locations and years showed that the null model SLY, SF did not fit the data. Lack of fit of the null model indicates that initial state was not sufficient for predicting fate given the distribution of states over locations and years. Location, year and the interaction between location and year made significant contributions to explaining variation in state by fate transitions for both M cardinalis and M Iewisii (Table 2). Spatiotemporal Variation in Reproduction Table 3 gives the results of mixed model analysis of variance tests of the effects of year, position within the range, and population nested within range position on reproduction. For M cardinalis, position within the range affected flower number and marginally affected fruit set. For M Iewisii, position within the range affected fruit set and seed number per fruit. Year affected M cardinalis seed number but did not affect M Iewisii reproduction. The effect of position within the range did not depend on year, as indicated by non-significant year by range position interactions for both species. Population and population by year interactions affected some reproductive variables for both species, but in general, between-population variation at a given range position did not overwhelm differences in reproduction between central and marginal areas of the elevation range. Mimulus cardinalis displayed reduced fecundity at the low elevation range center compared to the mid elevation range margin due primarily to reduced flower number per reproductive plant and reduced fruit set (Figure 2a, c, e). Mimulus 22 :8 $.80 SN: 3.2: tch N an; .03 000.0 00 000002 .00 00.00 000000000 00 0000. ...ta: $0: 0.2 00.0 ...Lefi :tfism 0| 3 8000 ..5 000-022 2...: 00.3. ~05 0: __ 00 ”5 5 at 5.80 00.00 30: $000 .00 00-0 .05 000.0 00 000» .00 000.00 0000000000 ”m 0000. 0.1.203 mwdaN ****Nm.wa 00006000 0:...ucém— 0| mg ANS $.03 mudN 0.00: 00.03 0N Ea >4 8% omdoc o_.MN_ 30-3 mod?” mm #0 £4 000.0 00 000003 .00 000,000 000008 ”N 0000- ......05 5.02 m20:: $30.00 5.00% .el 0; Ga 3.80 SN: $.80 EdoN 0N an; >0 030 8.80 2.000 0200 8.0:... mm 0.0 .5 000.0 00 000» .00 000.00 000.000 ”0 00.00. 0.002.000.0000 .3 _00000 003-0000 000000-000 .000000-000 £0002 G00 000m 00000 090.0 =08m .00 003-00000. 000000m D N .8030 0005 0000000 0000 00.0 mod .00 0000 00000 0 00b 0005-0300 0 000000000 00 000000000 000000000m 000000000 00000 0000000me 0000000 000.0000 00 m00_0> 000000005 000 00 000000000 00000 00000.0 03000.0 000 00000000000 00000-0032 :0 .00 00000000 000 000000 0000000000 00 00 00000000 000 0000 0000 E00000 000000000000 00.0 00000>000 030:8 000000 Z .0000» 000 0000002 w00000 0.00000 5 0000000000 000 00 0000000000 .0300 0000000000 0000-3000; 000 00 000» 000 000000— .00 0000.00 000 .00 _00000 003-0000 000 m0>0w 000—0 0w000 0000 .00 $000.00 000 w0000000m .m 000—0 0w000 0000 000 .0 .0000 00 .> .000» 000 J .000002 .0 00000.00 05 .00 0003000 00000004 .N 0300- 23 1:000 00.000 1.10 03.: :00 00.000 :30: 00.000 0|0|00 03: 0000 00.000 00.1%th :0.mcm Amt ondmm 0000 00.000 :10: 00-000 :00 00.000 800 00.000 :3?3 E c 000 00.000 0 .0 .0 *Ath mmdwN A03 CM. 3— :t00000 ...0-000000 5.00.00 0 9 00300 .5 0.0.00 00.0.01 0|$0 0|0 __ 0.0 .00 5 00.000 00.000 00.000 00 00> .5 03.0 G0 00000000— .00 000.000 30003000600 #0 60H. 00.: 3:00.00 $100.00 0 3 5:00 000 00.00: 0.0100 lawm- fl __ 0.0 0.0 5 00.00 00.00 00.000 00 r: .5 03.0 G0 000% .00 woo-0.00 3:033:00 ”m amou- ..:..:cm.ha :i—édi 000.00%.50 a n: Imiml 00.00 . 00. .00. ..0 .00 r: 5 00.000 00.0: 00.000 00 0.0 .5 00mm .00 0.0000000— .00 woo-0.00 ~000me08 ”N awe-H mzood 00w.N0 00......decu c 00> 00.000 00.000 00.000 0|0 0000 5 00.000 00.0: 00.000 00 00 .5 0000 00 0000 .00 000.000 00009000 ”0 000.0. 00003-030 0200.00 :05... 3:00.000 00 =0 00300 05 0.. 0 - 0. . . - 0- a 00.00 00.00 00.000 00 0.0 .0.» .5 000.0 0.050 0500000— 4000» .00 050000.505 0903-00.05 um ”Eu-H ****~m.ma :1.th0 ****mh.Nm~ o A> 0033 ”—4 00.00 00.3. 00.000 0.0. __ ..0-0 ..5 5 00.080 0 00000 00300 .00 24 .nim ...bwm .>Am 338 33.58 05 83w .mowfim =w 8>o 3883 555 _F .mu—m mrqm .258 $27.39* 05 83w .mowfim =a ~26 3683 can? . .gm .>Am 358 33.58 05 mozw .mowfim =m 83 3883 can? 0 .mm .>Am $on 33.38 05 83w .momfim =m ~26 vogm 555 H Sood v m1: .256 v mi... .56 v 9.... “mod v m... .36 v m v 3.3 .36 A mmz Ammmaég 31%va 2.3— mzmmdm :mwdm iihaém w _ 2m 533 “S: o o c o ol a $3 N3“: mmdm mw.wm 36m M: ”3 an; S4 8mm 98 20:82 .39» mo 830885 $3-025 ”m amok .3283 N 2an 25 Table 3. Results of mixed model analysis of variance testing the effects of range position, year, and population within range position on reproduction. Random effects denoted by ‘[R]’. Flower number log-transformed and fruit set (proportion of flowers maturing fruit) arcsine square-root transformed prior to analysis. F -tests for fixed effects constructed by SAS MIXED procedure, with denominator degrees of freedom obtained from the Satterthwaite approximation and indicated in parentheses below each F-value. x2 values for random effects from likelihood ratio tests. Values in bold remain significant afier sequential Bonferroni adjustment to maintain table-wide type I error of 0.05 for each species (Rice 1989). Response variable Species Source df Flowers/ plant Fruit set Seeds/ fruit M. cardinalis Position 1 2326*” 15.891‘ 2.34 (9.5) (1.96) (1.89) Year 3 2.71 1.22 12.40" * * (9.09) (5.62) (107) Year*Position 3 0.47 0.05 — (9.09) (5.62) Pop(Position) l 0.00 1.40 7.30” [R] Pop*Year(Position) 1 311' 31.20"" — [R] M. Iewisii Position 1 4.06 17.43" 91.8"" (1.65) (7.12) (138) Year 3 0.04 2.44 1.72 (4.25) (7.11) (138) Year*Position 3 0.14 2.03 — (4.25) (7.11) Pop(Position) 1 0.60 0.00 0.00 [R] Pop*Year(Position) l 5.20* 15.70*** [R] ‘l' 0.05 < P < 0.10, * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001. 26 Iewisii displayed highest fecundity at the high elevation range center and reduced fecundity at the mid elevation range margin due to lower fruit set and an approximately two-fold reduction in seed number per fruit (Figure 2b, d, 1). Seed Dormancy In 2003, 2291 seedlings emerged at May Lake from an initial total of 21,600 seeds placed at five seed stations in 2002, giving a germination percentage of 10.6%. At Buck Meadows, 38 seedlings emerged in 2003 out of an initial 2500 seeds at two stations, giving a germination percentage of 1.5%. Both germination estimates are corrected for seedlings emerging from “no seed” controls (May Lake: 17, Buck Meadows: 1). In 2004, 63 additional seedlings at May Lake and 7 seedlings at Buck Meadows emerged from dormant seed treatments, after correcting for seedlings in “no seed” controls (May Lake: 32, Buck Meadows: 0). Based on these observed germination rates and following the calculations of Horvitz and Schemske (1995), the estimated percentage survival of seeds was 19.9% for M. cardinalis and 13.4% for M. Iewisii (Appendix B). Although both species displayed similar overall seed survival, M. Iewisii seeds were more likely to germinate than to become dormant, whereas M. cardinalis seeds were more likely to become dormant than to germinate (Appendix B). Projection Matrix Analyses Lambda values ranged from 0.47 to 1.16 for M. cardinalis and from 0.68 to 1.33 for M. Iewisii (Figure 3; Appendix C). For M. cardinalis, lambdas at the low elevation range center were significantly lower than lambdas at the mid elevation range margin (Figure 3). The 95% confidence intervals for M. cardinalis low elevation lambdas never overlapped one, the value for stable population size, except at Buck Meadows from 27 M cardinalis M Iewisii ’5 Z 5 3 .2 LL. 2 1.0 - C 1.0 ~ D g H O 8 0 8 ~ Eé‘ ' ' a on 0.6 — 0.6 — 8 E g 0.4 - 0.4 — 0 0.2 - 0.2 — 5: , ll) 0.0 - . - . 0.0 . BU RP WA CA ML WF WA CA 2500 — E 2500 ‘ F E 2000 — . 2000 - 2 1500 - ‘ 1500 g 1000 - 1000 g 500 ‘1) 2 o BU RP WA CA Center Margin Center Margin Population Population Figure 2. Spatiotemporal variation in reproduction. a) M. cardinalis flower number, b) M. Iewisii flower number, 0) M. cardinalis fruit set, d) M. Iewisii fruit set, e) M. cardinalis seed number per fruit, and f) M. Iewisii seed number per fruit. Data presented are means + SE. Location abbreviations as in Figure 1. Year abbreviations as follows: 00 = 2000, 01 = 2001, 02 = 2002, 03 = 2003. 28 2000-2001. In contrast, 95% confidence intervals for all M. cardinalis mid elevation lambdas overlapped or exceeded one except at Wawona from 2002-2003. For M Iewisii, lambdas at the high elevation range center were significantly higher than lambdas at the mid elevation range margin. However, most 95% confidence intervals at one marginal location (Carlon) overlapped one, whereas 95% confidence intervals at the second marginal location (Wawona) did not. At high elevation, the 95% confidence intervals for all lambdas overlapped one except at May Lake from 2002-2003. For M. cardinalis, significant temporal (among-year) variation in lambda was detected at one central (Rainbow Pool) and one marginal (Wawona) location (Figure 3). For M. Iewisii, significant temporal variation in lambda was detected at all locations except for Wawona (Figure 3). . 2. 2 O A) M. cardinalis 0 B) M. Iewisii _ 00-01 1.5 ~— 01-02 1.5 « — 02-03 I: pooled 1.0 7 . t l 1.0 ‘ 0.5 ~ 0.5 ~ 71. with bias-corrected 95% Cl ' L .4. 0.0 - “1 0.0 ~ ' " BU RP‘ WC’" SC MU" WF‘" WL SL" 830 833 1208 1320 m 2690 2750 1208 1320 m Range center Range margin ‘ Range center Range margin Figure 3. Asymptotic population growth rates (A) for each location and transition interval and for pooled location matrices. Vertical bars indicate bias-corrected 95% confidence intervals (Caswell 2001). Asterisks indicate significant among-year variation within a location based on randomization tests (* P<0.05, ** P<0.01, *** P<0.001, **** P<0.0001). Pooled ,1 not sharing letters differ significantly from one another (after sequential Bonferroni adjustment) based on randomization tests. Abbreviations as in Figures 1 and 2. 29 Because separate estimates of dormancy from central and marginal populations were not available, I varied the dormancy component of the seed-to-seed transition by i 50%. Decreasing seed dormancy by 50% decreased lambdas by 0.9 — 1.9% for M. cardinalis and by 0.03 — 0.2% for M. Iewisii, and increasing seed dormancy by 50% increased lambdas by 1.3 — 3.0% for M cardinalis and by 0.04 — 0.2% for M. Iewisii. For both species, lambdas at all locations responded similarly to increases or decreases in seed dormancy, and the magnitude of change in lambda due to variation in the dormancy transition was not sufficient to erase differences in lambdas between central and marginal populations. Transition matrices and sensitivity matrices are given in Appendix C. For both species at all locations and for all transition intervals, lambda was most sensitive to perturbations in transitions from seeds to vegetative stage classes, particularly from seeds to the reproductive stage class (Appendix C). Lambdas of both species were also sensitive to perturbations in transitions to the reproductive stage class from vegetative stages. Bias-corrected 95% confidence intervals were broadly overlapping among transition intervals and locations, indicating that all location-year matrices had similar sensitivity structure (data not shown). Life Table Response Experiment LTRE analysis confirmed that, for M. cardinalis, sites at the range center had a negative effect on lambda whereas sites at the range margin had a positive effect on lambda, where the overall effect of a particular treatment level is estimated by summing the contribution of each matrix element to variation in lambda (Figure 4a). For M Iewisii, on the other hand, sites at the range center had a positive effect on lambda and 30 .N e3 _ 353 a a aeasefia. doom =36er mfitefim eeaeeudoe Xena veweeboeémfi 88:3: ween _we_te> .NBEE E aerate.» 3 EeEBe xEmE :35wa :eme .«o acceptance e5 wcmgm .3 deem—~50 Seebm ..w :0 use» use seamen: mo c2888.: e5 98 .3285 8:655 bane» .8382 we Beebe 5S, Amdhd ecefiteqxe encamee e38 £5 .v eSwE Sehéoumeoq Be> seaweed <0 <3 ”:5 1:2 5sz e850 <0 <3 m3 1:2 moumd No-8 Edd vd- . vd- . L ed- Z s- -2- B B -2- > ed _P w 1.] o.o , flu ed w .W U .I. No - No S - Ne w. m m. m we - vo - vo Wu 1 ed m r ed Q T ed kw Ewen: eeeceU mo-No no-5 ado <0 <3 mm Dm vd- _ . . vd- . . vd- Ne- - mo- - No- ed 3 FT. JD ed {SB 3 (suopnqtnuoo 1112);} snowman w Nd 1 Nd I Nd vd u vd . vd 0 ed m . ed < 1 ed Beebe 55223:— . Beebe 23> mueebe .8339— 31 sites at the range margin had a negative effect on lambda (Figure 4d). For both species, yearly transition interval had a much smaller effect on lambda than did location. For M cardinalis, 2001-2002 had a positive effect on lambdas, and 2002-2003 had a negative effect on lambdas (Figure 4b). For M Iewisii, year effects did not differ from zero (Figure 4e). The interaction of location and year affected M. cardinalis lambdas at Wawona and M Iewisii lambdas at all sites except Wawona. At Wawona, M cardinalis lambdas were significantly higher than expected based on the main effects of location and year in 2000-2001 and significantly lower than expected in 2002-2003 (Figure 4c). For M. Iewisii, lambdas in 2001-2002 were higher than expected at the range center and lower than expected at the range margin. The converse was true in 2002-2003 (Figure 4f). Several transitions made large contributions to spatial variation in lambda (Figure 5). For M cardinalis, fecundity transitions from large non-reproductive and reproductive individuals to the seed class and stasis in the reproductive class had large negative effects on range center lambdas and large positive effects on range margin lambdas. In contrast, recruitment from seed to the large non-reproductive class made a positive contribution to lambda at the range center. Contrary to M. cardinalis, fecundity transitions from large non-reproductive and reproductive classes to seeds and stasis in the large non-reproductive and reproductive classes negatively affected M Iewisii range margin lambdas and positively affected range center lambdas. A large positive contribution of recruitment from seed to the large non-reproductive class also partially offset the negative contributions of other M. Iewisii transitions at Carlon. The M. cardinalis year and location by year interaction effects (Figure 4b, c) 32 .e>oe=voaeenee .eZHesdoEee-co: ewe—2 u E .ezeenooaee-co: =25 n am deem u R 630:8 me 3233958“ mmfle emfim ._ 2:3.» 5 3 2233959“ 5:83 .SooN :95er £9.52: eeceocaoe Reno oeueebee -35 88MB: mad Remte> .N 3 count? 35% 9 mnemoaowncoe Bone—e 5.92: coming-w .m 8:me someone“:- :oEmSfi-r ameammwmmmmmmmmm eaammmmmmmmmmeww meWanWam-mflmflmm. meWquWanWmeW PM ri . . r p p . p r b r F . . . I . p p p . . . . > r r L M.°l 0 me me. <0 fill”. - Nd- m. <3 I m... n 1 I ~00 m. U . s - co m- . A . _o m. me- No - No m. Ewan: emcee mfiifi we EweeE emcee 35528 .3 m. . .o me n v- amammmmmmmmmwmmm eeammmmwmmmmmmmm quWmm-mmmw-mmmwimm. mw-mem-mmmmmmmnlemm ..an » p b . t . L . . i . . . I i p . r L . . f h p L f » n.0l 0 mo m U . nu . m3 - No- Ad - No- m. A: I Dm— I m- . I i I u. - _ o _o O t w - co __ co m. A - do - _o m. we.- - Nd - Nd m. .2823 emcee 53$ .3 eeecee emcee QGEBLS .3 u. mo mo W 33 were due to contributions from the same fecundity and stasis transitions that gave rise to the location effect (data not shown). For M. Iewisii, however, location by year interaction effects arose primarily from spatiotemporal variation in the contribution of recruitment to the large non-reproductive stage class, a transition with high sensitivity. From 2000-2001 and 2001-2002, recruitment at the range center was high and recruitment at the range margin Carlon site was low; however, this difference was reversed from 2002-2003, when recruitment of large non-reproductive plants at Carlon was high and recruitment at the range center was low (data not shown). T ransects All locations selected for detailed demographic study fell within the range of plant densities observed at similar elevations (Appendix A). Density of M cardinalis small non-reproductive plants increased with elevation, although this difference did not remain significant after sequential Bonferroni adjustment (Table 4). Elevation did not predict density of other M. cardinalis stage classes or of any M. Iewisii stage classes, nor did elevation predict temporal variation in stage class density of either species (Table 4). DISCUSSION Variation in Vital Rates Observations of population vital rates demonstrate variation in performance across the elevation ranges of M. cardinalis and M. Iewisii. For M. cardinalis, survival of reproductive plants was higher at the range margin than at the range center, whereas for M Iewisii, survival of this stage class was higher at the range center than the range margin. Log-linear analysis of vegetative stage class transitions revealed significant 34 Table 4. Linear regressions of stage class density (2001-2003 mean number of plants per m2) and temporal variation in stage class density (coefficient of variation, CV, in 2001-2003 density) versus elevation along 50-200 m A transects. Afier sequential Bonferroni correction to maintain a table-wide type I error rate of 0.05, no regression coefficients differed from zero. Species Dependent variable N b SE(b) t P M. cardinalis Sm. non-repro. 18 0.00062 0.00023 2.68 0.02 Lg. non-repro. 18 0.00001 0.00001 1.73 0.10 Repro. density 1 8 0.00005 0.00006 0.84 0.41 CV (sm.non-repro.) 18 -0.02212 0.04431 -0.50 0.63 CV (lg. non-repro.) 18 -0.01918 0.05017 -0.38 0.71 CV (repro.) 18 -0.00898 0.03624 -0.25 0.81 M. Iewisii Sm. non-reprod. 20 0.00003 0.00007 0.40 0.69 Lg. non-reprod. 20 0.00000 0.00000 0.68 0.51 Reprod. 20 0.00006 0.00003 1 .73 0. 10 CV (sm.non-reprod.) 20 -0.00931 0.01723 -0.54 0.60 CV (lg. non-reprod.) 20 0.02305 0.02070 1.11 0.28 CV (reprod.) 20 0.00191 0.02035 0.09 0.93 temporal and spatial variation in the fates of vegetative plants. Components of plant fecundity also displayed significant variation between central and marginal populations of both species. Fecundity of M. Iewisii was higher at its high elevation range center and lower at its mid elevation range margin. Reduction in plant fecundity at the range margin arose due to fewer flowers maturing fruit and an approximately two-fold reduction in seed number per fruit. It is unclear whether reduced fruit set and seed number per fruit resulted from physiological limitations on seed maturation or from 35 pollen limitation. F ecundity of M. cardinalis was higher at its mid elevation range margin and lower at its low elevation range center. Reproductive plants at the range center produced fewer flowers per stem and were of overall smaller size than reproductive plants at the range margin, resulting in fewer flowers per reproductive plant than at the range margin. Variation in Population Growth Rates The projection matrix summarizes how a particular environment affects the demographic parameters of a population. The asymptotic population growth rate, 11, is the rate at which the population would grow were the present environmental conditions to remain constant. Although the assumption of time invariance is almost certainly invalid, matrix projections remain extremely useful for summarizing the effects of different environmental conditions on projected population growth rates and population structure. Because In A = r, the instantaneous grth rate, ,1 may also be interpreted as the average fitness of the population in the given environment (Fisher 1930; Charlesworth 1980; Caswell 2001). In this study, matrix projections revealed large differences in A of central and marginal populations for both M. cardinalis and M. Iewisii. Projected population growth rates of M. Iewisii were highest at the high elevation range center and reduced at the mid elevation range margin. Projected population grth rates of M. cardinalis were highest at the mid elevation range margin and greatly reduced at the low elevation range center. Asymptotic projections were similar to observed year-to-year changes in population size. For example, the observed 2002-2003 population growth rate at Rainbow Pool was 0.4769, as compared to the asymptotic population growth rate of 0.4724. 36 Some temporal variation in population growth rates was also detected, but inspection of regional climate records does not reveal a clear relationship with variation in climatic variables such as precipitation or temperature. Temporal variation in 2. observed during this four-year window may be due to within-site processes such as frequency of tree falls than climatic variation. Although statistically significant variation among years was detected at most locations with randomization tests, in general the magnitude of temporal variation was smaller than spatial variation during the study period, a finding supported by results from LTRE analysis. However, temporal environmental variation can play an important role in the population dynamics of riparian plant species (Menges 1990; Lytle and Merritt 2004), and it remains possible that temporal variation acting over a longer time scale or at irregular intervals has important consequences for Mimulus population dynamics. Population grth rates of M. Iewisii fit the expectation that central populations have high fitness and marginal populations have reduced fitness. Population growth rates of .M. cardinalis, on the other hand, displayed the opposite pattern. The strikingly low As observed for M. cardinalis at its range center contrast with results from a reciprocal transplant experiment in which M. cardinalis and M. Iewisii were grown at 415, 1400, 2395 and 3010 m. In reciprocal transplant gardens, M. cardinalis and M. Iewisii displayed the greatest average fitness at their respective low (415 m) and high (2395 m) elevation range centers, and reduced fitness at the mid elevation range margin (1400 m; Chapter 2). Reciprocal transplants and demographic observations have distinct advantages and disadvantages, and together, the two methods provide complementary information about how performance varies across species’ ranges. By definition, 37 observations of extant populations cannot determine fitness levels beyond present range boundaries. Reciprocal transplant experiments are a powerful way to test for fitness variation both within and beyond present range limits, and the purpose of the reciprocal transplant experiment was to examine the effects of macroclimatic variables within and beyond the species’ elevation ranges on components of fitness. To accomplish this, seedlings were grown in relatively uniform and favorable conditions (e.g., irrigated plots, minimal competition) to isolate the effects of climate on performance. However, because experimental gardens were established with seedlings, seed to seedling transitions were not observed. Observations of natural populations, on the other hand, integrate performance throughout the life cycle over all underlying, but often unknown, environmental variables. One possible explanation for low is at the M. cardinalis range center and the M. Iewisii range margin is that downstream populations are demographic sinks maintained by immigration from upstream populations. Little is known about mechanisms of dispersal of M. cardinalis and M. Iewisii seeds. Because both species occur in riparian habitats, it is possible that seed dispersal via downstream currents provides a mechanism for primarily unidirectional long-distance dispersal among populations, as has been demonstrated for M. guttatus (Waser et al. 1982). Alternatively, temporal variation, particularly related to flood cycles, may operate over a longer time scale than the duration of this study and may have different effects on low versus mid elevation populations of M. cardinalis, leaving open the possibility that low elevation populations experience better “good” years than mid- elevation populations. Periodic floods may cause boom-bust cycles of mortality, bursts 38 of recruitment, and subsequent population attrition (Lytle and Merritt 2004). Low elevation populations may undergo greater variation following floods due to increased magnitude of floods on larger waterways at low elevation and/or to greater potential growth and fecundity of plants at low elevation in wet years. Examination of regional stream flow records (http://waterdata.usgs.gov/ca/nwis/nwis) confirms that flood magnitudes, both in absolute terms and in deviation from average peak flows, increase at lower elevations as catchment area increases. This hypothesis is consistent with the observation that, at low elevation, plants were recorded high on riverbanks and relatively distant from water at the beginning of the study, only three years after the largest recorded flood in the region (January 1997). Populations have since retreated to areas closer to water. This hypothesis is also consistent with plant performance in irrigated reciprocal transplant gardens, in which plant performance was measured under optimal conditions, and M. cardinalis exhibited greatest growth and reproduction at the low elevation range center (Chapter 2). A similar interaction between temporal variation and range position may also be possible for M. Iewisii, although it is likely to be of limited extent due to smaller flood magnitude at mid and high elevations and limited grth potential of plants at mid elevations (Chapter 2). Further studies of both seed dispersal and spatiotemporal variation in population dynamics are necessary. Only a handful of studies have examined the demography of geographically central and marginal native plant populations, each finding unique patterns of variation between central and marginal locations. Nantel and Gagnon (1999) studied two clonal plant species, Helianthus divaricatus and Rims aromatica, and found that all populations exhibited high growth rates at least some of the time, but that northern 39 peripheral populations exhibited greater temporal variation in population growth rates than more centrally located populations. In a study of the annual grass Hordeum spontaneum along an aridity gradient from the center to the margin of its range, Volis et al. (2004) reported greater population grth rates in central populations in most years. However, local adaptation of seed dormancy traits in marginal desert populations ensured population persistence through drought periods. Finally, Stokes et al. (2004) examined congeneric shrubs, Ulex gallii and U. minor, whose parapatric distributions they hypothesized were limited by competition, but found that both species exhibited greatest population growth in marginal, sympatric areas. The present study also examined the population dynamics of closely related congeners in marginal areas of sympatry, but it was not designed to estimate the effects of competition between M. cardinalis and M. Iewisii on vital rates and population growth. It is interesting to note that marginal locations at mid elevation had negative effects on M. Iewisii As and positive effects on M. cardinalis is, but from this study it is not clear to what extent this is due to competitive superiority of M. cardinalis versus adaptation of M. Iewisii to high elevation environments. However, even with minimal competition, M. Iewisii exhibits low fitness in reciprocal transplant gardens at middle and low elevations (Chapter 2) as well as in temperature regimes characteristic of low elevation (Chapter 3), suggesting that adaptation, or lack thereof, to the abiotic environment plays an important role in the performance of M. Iewisii at its range margin. Contribution of Life History Transitions to Variation in Population Growth Rates 40 Analysis of transition matrix data as a life table response experiment revealed several important life history transitions that contributed to differences in lambda between central and marginal populations. Transitions from large non-reproductive and reproductive plants to the seed class and stasis in the reproductive class made the largest contributions to spatial differences in lambda. These transitions had only low to moderate sensitivities, and sensitivity values were largely similar across all locations, indicating that differences in projected population growth rates resulted mainly from observed differences in transition matrix parameters. At the mid elevation range margin, M cardinalis was more likely to become or remain reproductive and made more seeds per individual than at low elevations, and these differences in vital rates contributed to the observed differences in key transition matrix elements. Similar patterns of difference were observed for M. Iewisii at the high elevation range center versus the mid elevation range margin. Variation in local population density Local population density is often used as an indicator of the degree to which a particular environment meets the niche requirements of a species (Brown et al. 1995). Many studies have concluded that abundance does in fact decrease towards range margins (McClure and Price 1976; Hengeveld and Haeck 1982; Huff and Wu 1992; Svensson 1992; Telleria and Santos 1993; Brown et al. 1995), but many others have not (Blackburn et al. 1999; Perez-Tris et al. 2000), and a recent review determined that fewer than half of all such studies found support for this generalization (Sagarin and Gaines 2002). The present study finds no clear relationship between population mean 41 fitness, as measured by A, and local population density, despite differences in population growth rates between central and marginal areas of the elevation range. In sum, this study demonstrates that central and marginal populations of both M cardinalis and M. Iewisii differ in vegetative stage class transitions and fecundity, and that these differences in vital rates contribute to substantial spatial variation in population grth rates. Continued study of spatiotemporal variation in population dynamics, in combination with estimates of dispersal between central and marginal populations, will improve our understanding of species’ distribution limits. 42 CHAPTER 2 Variation in fitness within and beyond Mimulus cardinalis and M. Iewisii elevation ranges Abstract—Every species occupies a limited geographic area, but it remains unclear why traits that limit distribution do not evolve to allow range expansion. Hypotheses for the evolutionary stability of geographic ranges assume that species are maladapted at the range boundary and unfit beyond the current range, but this assumption has rarely been tested. To examine how fitness varies across species ranges, I reciprocally transplanted two species of monkeyflowers, Mimulus cardinalis and M. Iewisii, within and beyond their present elevation ranges. I used individuals of known parentage from populations collected across the elevation ranges of both species to examine whether populations are adapted to position within the range. For both species I found the greatest average fitness at elevations central within the range, reduced fitness at the range margin, and zero or near-zero fitness when transplanted beyond their present elevation range limits. However, the underlying causes of fitness variation differed between the species. At high elevations beyond its range, M. cardinalis displayed reduced growth and fecundity, whereas at low elevations M. Iewisii experienced high mortality. Weak differences in performance were observed among populations within each species and these were not related to elevation of origin. Low fitness of both species at their range margin and weak differentiation among populations within each species suggest that adaptation to the environment at and beyond the range margin is hindered, illustrating that range margins provide an interesting system in which to study limits to adaptation. 43 Key words: range limit, evolution of species’ distributions, elevation gradient, reciprocal transplant, survivorship analysis Every species occupies a restricted geographic area. In some cases, geographic ranges stop at an obvious barrier, such as a land — water interface. However, more frequently, ranges end at “seemingly arbitrary” points in space (Kirkpatrick and Barton 1997). Historically, ecologists and biogeographers have correlated range boundaries with climate to identify environmental determinants of range boundaries (Griggs 1914; Good 1931; Dahl 1951). Subsequent analyses have shown that range limits are associated with abiotic variables such as temperature or precipitation (Root 1988a; Cumming 2002), biotic factors such as competitors (Terborgh and Weske 1975; Bullock et a1. 2000) or complex interactions between biotic and abiotic variables (Randall 1982; Taniguchi and Nakano 2000). Even a mechanistic understanding of the relationship between environmental variables and distribution limits presents an evolutionary conundrum. Natural selection should continually improve adaptation at a range boundary and thus overcome current geographic limits, causing species’ ranges to “grow by a process of annual accretion like the rings of a tree” (Mayr 1963). Several hypotheses for the evolutionary stability of range limits propose that populations at range boundaries do not have sufficient genetic variation to respond to natural selection (Bradshaw and McNeilly 1991; Hoffman and Blows 1994; Gaston 2003). Other hypotheses focus on other factors that may prevent populations from adapting to the environment at the range margin, such as genetic trade-offs among fitness-related traits in the marginal environment (Antonovics 44 1976), genetic trade-offs between fitness in central and border environments (Holt 2003), or gene flow from populations adapted to the range center (Haldane 1956; Garcia-Ramos and Kirkpatrick 1997; Kirkpatrick and Barton 1997). These hypotheses are not necessarily mutually exclusive, and may act synergistically to constrain range expansion. All of the above hypotheses are united by the assumption that populations are maladapted at a range boundary and unfit beyond the current range. A corollary of this generalization is that concomitant environmental changes impose selection for local adaptation to the range edge. Surprisingly, these assumptions have rarely been directly tested. Indirect evidence for a decline in fitness with distance from the range center is provided by the observation that, in some species, numerical abundance decreases with distance from the range center, presumably in response to an increasingly unfavorable environment (Brown 1984; Brown et al. 1996; Sagarin and Gaines 2002). Other indirect evidence for changes in fitness across species ranges comes from studies of fluctuating asymmetry. Developmental instability may increase when organisms are under genetic or environmental stress, as is predicted for individuals at range boundaries, and several studies of fluctuating asymmetry have found that populations at range boundaries do have higher levels of fluctuating asymmetry than central populations (Mallet 1995; Carbonell and Telleria 1998; Gonzalez-Guzman and Mehlman 2001). A more critical test for reduced fitness in marginal populations involves direct observation of fitness components across species ranges. Such studies have often found lower survival of certain life history stages or reduced fecundity at the range margin 45 relative to the range center (Marshall 1968; Pigott and Huntley 1981; McKee and Richards 1996; Garcia et al. 2000; Hennenberg and Bruelheide 2003). Unfortunately, the demographic consequences for such reductions in fitness are generally unclear. Perhaps the biggest stumbling block to observations of fitness variation, however, is that by definition, observations of extant populations cannot determine fitness levels beyond present range boundaries (Woodward 1990). Reciprocal transplant experiments are a powerful way to test for fitness variation both within and beyond present range limits as well as the presence of genetically based local adaptation (e.g., Schemske 1984; Stanton and Galen 1997; Verhoeven et a1. 2004). Although many classic studies used reciprocal transplants between areas within species ranges (Turesson 1922; Clausen et al. 1940), few have transplanted individuals beyond the range (Gaston 2003). I used reciprocal transplants to evaluate population and geographic variation in fitness for sister species of monkeyflower, Mimulus cardinalis and M. Iewisii (Phrymaceae) across their elevation ranges in California, USA. The study of closely related species with distinct distributions offers a conceptual advantage for the investigation of range limits. In a comparison of central versus border populations of a single species, one could never reject the possibility that border populations have not yet acquired the right mutation(s) to extend the border. In a comparison of parapatric sister species partitioning an environmental gradient, evolution from the common ancestor toward each species’ native environment has already occurred, and the question of interest is what causes and constrains adaptation to different ends of the gradient. 46 Mimulus cardinalis and M Iewisii have been the subject of ecological and genetic studies for several decades and have many properties that make them ideal research subjects, including high seed number, high germination rates, and low transplant mortality (Vickery 1967; Hiesey et al. 1971; Vickery 1978; Bradshaw et al. 1998; Bradshaw and Schemske 2003; Ramsey et al. 2003). Pioneering studies of M. cardinalis and M. Iewisii by Hiesey et al. (1971) revealed variation in performance across elevation, with M cardinalis displaying low survival and reproduction at high elevation and M Iewisii displaying low survival and grth in a coastal climate. Unfortunately, several features of this study limit its usefulness for drawing definitive conclusions about variation in fitness versus elevation. First, populations were collected throughout the geographic ranges of both species from Washington to Baja California, but transplanted at only three sites (Stanford, elev. 30; Mather, elev. 1400; and Timberline, elev. 3050) along a narrow elevation transect in northern California. The wide latitudinal and longitudinal distances that separated most populations from the transplant sites are not easily separated from the effects of adaptation to elevation. Although the authors found significant population differentiation within each species (e. g., between coastal Californian and montane Arizonan M cardinalis), regional and subspecies differences are not easily separated from differences related to elevation alone. Second, the use of vegetatively propagated clones eliminated information about the performance of early life history stages that may experience strong selection and be critical for population establishment (Travis 1994; Caswell et al. 2003; Davis et al. 2003; Lee et al. 2003; Zacherl et al. 2003). Finally, the low elevation transplant station 47 at Stanford (30 m) potentially conflated the effects of low elevation with a maritime climate. 1 used reciprocal transplants within and beyond the elevation ranges of M cardinalis and M Iewisii to examine how survival, growth and reproduction of each species change with elevation. I used individuals of known parentage from populations collected across the elevation ranges of both species to examine whether populations are adapted to their position within the range. Specifically, I asked 1) How do fitness components change from the center to the edge of ranges and beyond? and 2) Are populations locally adapted within their range? MATERIALS AND METHODS Study System Mimulus cardinalis and M Iewisii (Phrymaceae) are rhizomatous perennial herbs that grow along seeps and stream banks in western North America. The species are self-compatible and animal pollinated (Hiesey et al. 1971; Schemske and Bradshaw 1999). Mimulus cardinalis occurs from southern Oregon to northern Baja California and from the coast of California inland to Arizona and Nevada. Mimulus Iewisii is composed of two races, a northern form occurring from southern coastal Alaska to southern Oregon and eastward to the Rocky Mountains, and a southern form, occurring primarily in the Sierra Nevada Mountains of California (Hiesey et al. 1971; Hickman 1993; Beardsley et al. 2003). The two races are partially incompatible, and recent phylogenetic analysis suggests that the two races are sister to one another and together are sister to M cardinalis (Beardsley et al. 2003). Here I study only the Sierran form of M Iewisii. 48 Mimulus cardinalis and M Iewisii segregate by elevation, with M cardinalis occurring from sea level to 2400 m and M Iewisii occurring from 1200 m to 3100 m (Hiesey et a1. 1971; Hickman 1993). In the Yosemite National Park region where this research was conducted, the species co-occur on larger watercourses between 1200 and 1500 m elevation (A. Angert, unpub. data). Although the published Californian distributions of M cardinalis and M Iewisii extend to 2400 and 3100 m, respectively, repeated attempts to locate extant M cardinalis populations above 1500 m and M Iewisii populations above 2900 m in the Yosemite region were unsuccessful. Therefore, I consider 1200 - 1500 m to be the shared mid-elevation distribution limit for both species and the western longitudinal distribution limit for M Iewisii. Genetic Material: Population Collection and Crossing Design Seeds from eight plants from each of six populations per species were collected in September 1999 along an elevation gradient from 590 m to 2750 m between 37.49 and 37.96° N latitude (Appendix A). One plant from each field-collected family was grown to flowering in the University of Washington greenhouse under standard greenhouse conditions. The eight plants from each population were crossed with one another in a partial diallel mating design (one per population, for a total of 12 partial diallels), where each plant served as sire and dam twice with no self- or reciprocal pollinations. Pollinations were performed by collecting all of the pollen from one flower with a flat toothpick and fully saturating the stigma of one flower. Seeds from four pollinations per full-sib family were pooled. This crossing design was intended to provide a genetically variable, outcrossed seed pool for reciprocal transplants rather than to accurately estimate genetic variance components. Sire and dam effects were 49 included in statistical models to account for the possible correlation of error and non- independence of individual measurements due to their family structure. Reciprocal Transplant Methods Garden locations.— To examine how species’ performance varies across elevation ranges, I established experimental gardens along an elevation transect on the western slope of the Sierra Nevada Mountains. In June-July 2001, gardens were planted near Jamestown, California (37.917°N, 120.421°W; elev. 415 m), at Carnegie Institution of Washington field stations at Mather (37.886°N, 119.855°W; elev. 1400 m) and Timberline (37.962°N, 119.281 °W; elev. 3010 m) and at the White Wolf Ranger Station in Yosemite National Park (37.872’N, 119.651°W; elev. 2395 m). These gardens were chosen to represent elevations for each species that are central within the elevation range (415 m for M cardinalis, 2395 m for M Iewisii), at the range boundary (1400 m for both species, 3010 m for M Iewisii), and beyond the range boundary (2395 and 3010 m for M cardinalis, 415 m for M Iewisii) in the Yosemite region (Figure 6). :1 M Iewisii Sympatry - M cardinalis Jamestown Mather White Wolf Timberline 415m |400m 2395m 3010m l Figure 6. Schematic transect of the central Sierra Nevada Mountains, California, showing M Iewisii and M cardinalis elevation ranges and placement of reciprocal transplant gardens, after Clausen et al. (1948). 50 Garden conditions.— Due to the tiny seed size and particular microhabitat requirements for germination of M cardinalis and M Iewisii, experimental gardens were established with seedlings. Seeds from partial diallel crosses were sown in flats in the University of Washington greenhouse five weeks prior to transport to garden sites. The average age of transplanted seedlings was approximately three weeks after germination, corresponding closely to the size of plants observed in natural populations at the time of planting. Two seedlings from each firll-sib family were planted at 10-cm intervals in a randomized block design for a total of 384 seedlings per block (2 seedlings / family x 16 full-sib families / population x 6 populations / species x 2 species). During June-July 2001, seedlings were planted in 3 blocks at 415 m (N=1152), 4 blocks at 1400 m (N=1536), 4 blocks at 2395 m (N=1536), and 3 blocks at 3010 m (N=1152), for a total of 5376 seedlings across all four transplant sites. Garden plots were covered in landscape fabric and irrigated daily to mimic conditions in the species’ native riparian habitat and to standardize water treatments across environments. Soils assay.— I collected soil samples from each garden site and grew plants in these soils under uniformly favorable greenhouse conditions to determine if site differences in performance were due to the effects of soils as opposed to other environmental factors. I measured the performance of four populations per species, using a subset of four independent full-sib families per population from the partial diallel crosses. Plants were able to flower on all soil types in the greenhouse environment and there was no evidence of local adaptation to soil type, therefore, I conclude that differences in soil properties are not primarily responsible for differences in fitness across elevation and I do not consider soil type further. 51 Measurements.— To assess fitness within each garden, I measured survival, growth and reproduction. Plants grew at vastly different rates among gardens. At 1400, 2395 and 3010 m, plants grew slowly and rarely attained a size where larger plants spread via rhizomes into neighboring plants’ space. However, at 415 m, M cardinalis plants began to spread via rhizomes into neighbors’ space afier one growing season, making it difficult to separate individuals and track identity. For this reason, I truncated observations at 415 m after one year, when all M Iewisii individuals were dead and surviving M cardinalis were very large. Individuals transplanted in a large preliminary study at 41 S m displayed very low mortality and continued rapid growth during the second growing season, indicating that truncation afier one year does not bias the results (A. Angert, unpub. data). Survival was monitored from 2001 — 2002 at 415 m and from 2001 — 2003 at 1400, 2395 and 3010 m. Survival was recorded at approximately two-week intervals throughout each growing season. Growth and reproduction were measured for one growing season at 415 m and for two growing seasons at 1400, 2395 and 3010 m. To measure plant growth, I recorded the total stem number and length of all stems. Stem number and total stem length were strongly correlated (M cardinalis: R2=0.73, N=2065, P=<0.0001; M Iewisii: R2=0.72, N=1790, P=<0.0001). I present stem length data because they better describe overall plant size at high elevations, where plants often have only one stem but differ in stem length. Because permit restrictions prevented seed set at two transplant sites, I use flower number rather than seed number as a proxy for reproductive fitness. Flower number and fruit number measured from 2000 — 2004 in demographic census plots within natural central and border populations 52 are highly correlated (M. cardinalis: R2=0.97, N=1132, P<0.0001; M. Iewisii: R2=0.98, N=1064, P<0.0001), suggesting that cumulative flower number is a good approximation of total fitness. I estimated overall plant fitness, retaining zeros for plants that failed to flower or failed to survive, as the cumulative flower number over two growing seasons. I also summed year one and year two total stern length to estimate cumulative growth. For M cardinalis grown at 415 m, only first year measurements of stem length and flower number were available. To keep measures comparable across all sites, I annualized measures of growth and fitness and compared average annual stem length and average annual fitness. Comparisons of first year growth and fitness at all sites as well as cumulative growth and fitness with the 415 m site excluded produced similar results; I present comparisons of annual averages for brevity. Statistical Analysis To examine fitness variation across species’ elevation ranges, I analyzed the relationships between transplant site and the fitness components of survival and growth and between transplant site and average annual fitness. Too few individuals remained alive and flowering beyond their ranges to allow analysis of flower number for surviving plants. To determine whether populations are adapted to range position, I analyzed the relationships between population origin and performance within each transplant site. For all variables, I conducted separate analyses for each species. All analyses were performed in SAS, version 8.2 (SAS Institute, Inc., Cary, NC). Survivorship.— I used accelerated failure time models to test for differences among sites and populations in patterns of survivorship. Accelerated failure time 53 models assume that factors affect failure time (e.g., time to mortality) multiplicatively, shifting the time periods when failures occur (see Fox 2001 for a general discussion of failure time analyses). For this study, accelerated failure time models were biologically appropriate because environmental differences among transplant treatments were expected to shifi the distribution of time to failure (Jones and Sharitz 1998; Keith 2002; Denham and Auld 2004). To apply the accelerated failure time model, I used PROC LIFEREG with an underlying Weibull distribution of failure time (measured in days after transplantation). Survivorship was described using the function: so) = e'W’p, where the scale parameter ll. scales the model to a baseline rate of mortality, t is the time since transplantation, and p is a dimensionless shape parameter that describes change in failure hazard over time, such that when p < 1 hazard monotonically decreases with time and when p > 1 hazard monotonically increases with time (Dudycha and Tessier 1999; Fox 2001; Keith 2002). I also ran models using an alternative plausible distribution, the exponential, which is a special case of the Weibull with the shape parameter p = 1, indicating a constant risk of mortality (Fox 2001). The exponential distribution gave a significantly poorer fit to the data than the Weibull according to likelihood ratio tests (M cardinalis: 76:2889, P<0.0001, M Iewisii, x2=31.8, P<0.0001) but yielded qualitatively similar results, indicating that the results are robust to the underlying distribution. For each species, I fit models with fixed effects of site, population and their interaction. For each categorical variable, one level was arbitrarily chosen as the reference level and its regression coefficient was set to zero. Regression coefficients and significance of all other levels were determined relative to the 54 reference, but this did not reveal whether differences among non-reference levels existed. Multiple comparisons were necessary to examine differences among levels other than the reference. I constructed Z-tests for multiple comparisons from estimated regression coefficients and the asymptotic covariance matrix according to the methods of Fox (2001). Because effects act multiplicatively on failure time, regression coefficients less than zero can be interpreted as shrinking the time to failure relative to the reference level, whereas positive regression coefficients expand the expected time to failure relative to the reference (Dudycha and Tessier 1999). Standard statistical packages do not incorporate random effects in survival time analyses, so for these analyses I was not able to include sire, dam or block effects. Observations were right censored if the individual remained alive at the end of the observation period. Growth.—— To examine the relationship between growth and transplant site, I performed mixed model analysis of variance on log-transformed data with PROC MIXED, which uses the restricted maximum-likelihood method (REML) to estimate variance components. I tested for variation in average annual stem length with respect to transplant site, population of origin, sire within population of origin, dam within each population of origin, and all interactions. Models including random block effects failed to converge, so I excluded block from the analyses. For this and all subsequent models, I considered transplant site and population of origin as fixed effects and sire and darn as random effects. To evaluate the significance of fixed effects, I used Type III estimable functions, which tolerate unbalanced samples, with denominator degrees of freedom obtained by Satterthwaite’s approximation. Differences among levels of fixed effects were evaluated with Tukey-Kramer adjusted comparisons of least square means. I used 55 the PDMIX800 macro to convert pairwise differences between least square means to letter groupings, where means sharing the same letter code are not significantly different (Saxton 1998). I used likelihood-ratio tests (comparing each reduced model to the full model including all effects) to evaluate the significance of all random effects. Only two M Iewisii individuals remained alive for stem length measurements at 415 m, causing the full model containing all sites to contain many non-estimable parameters. To remedy this, I excluded the 415 m site from the M Iewisii stem length analysis. F itness.— I used mixed linear models to test for variation in average annual fitness with respect to transplant site, population of origin, sire within each population of origin, dam within each population of origin, and all interactions. The distribution of fitness was highly non-normal due to an excess of zeros and a long right tail. Examination of residuals in preliminary analyses revealed significant departures from parametric assumptions. Transformations only slightly improved the distribution of residuals. Therefore, I used two approaches to model annual fitness. First, I performed mixed model analysis of variance on log-transformed data with PROC MIXED as described for stem length above, with the exception that I first added 1/6 to each observation before log transformation (Kuehl 2000). Second, I used the GLIMMIX macro of PROC MIXED to fit generalized linear models, which are appropriate for a wider range of error structures than traditional linear models (Kuehl 2000). Generalized linear models extend traditional linear models in two key ways. First, they allow the distribution of the response variable to be any member of the exponential family of distributions (e.g., gamma, Poisson, binomial). Second, they relate the response variable to a set of linear predictor variables through a nonlinear link function (SAS 56 SASInstitute 1999). The GLIMMIX macro uses restricted/residual pseudo likelihood (REPL) estimation to fit a generalized linear model with random effects. I modeled variation in average annual fitness using a gamma distribution with a log link function, which is appropriate for positive, continuous data (SAS SASInstitute 1999; Juenger and Bergelson 2000). Observations were first transformed by adding one to each observation. I used Type III functions with denominator degrees of freedom obtained by Satterthwaite’s approximation to test the significance of fixed effects. To evaluate the significance of random effects, I used the covtest option to obtain Z-tests, which tested whether the Z-value of each effect (its variance parameter divided by its approximate standard error) was different from zero (J uenger and Bergelson 2000). Because results obtained from PROC MIXED and GLIMMIX did not differ qualitatively and because the data violated the assumptions of traditional linear analysis, I present only the results from GLIMMIX. Population variation.— To evaluate whether populations are adapted to their elevation of origin, I used two approaches. First, I examined population by site interactions in the analyses described above. A significant population by site effect indicates that populations differ in their response to elevation. If a significant population by site effect was found for failure time, I compared the confidence intervals of regression coefficient estimates to determine which population and site combinations were significantly different from one another. If a significant population by site effect was found for growth or fitness, I used Tukey-Kramer adjusted comparisons of least square means to determine which population and site combinations were significantly different from one another. Second, if populations are locally adapted to their elevation 57 of origin, then fitness should decrease as the difference between elevation of origin and transplant site elevation increases. For each transplant site, I examined the rank correlations of population average annual fitness with the absolute value of the difference between origin and transplant elevations using PROC CORR. Table 5. Analysis of accelerated failure-time models for survival time, using 1339 uncensored values and 1273 right-censored values for M cardinalis, 1339 uncensored values and 1073 right-censored values for M Iewisii, and a Weibull distribution. Species Variable df Estimate SE )8 P M cardinalis Site 3 350.17 <0.0001 (Jamestown, 415 m) 1 -0.6896 0.1298 28.24 <0.0001 (Mather, 1400 m) 1 0.4662 0.1143 16.63 <0.0001 (White Wolf, 2395 m) 0 0 0 (Timberline, 3010 m) 1 0.2233 0.1047 4.55 0.0330 Population 5 6.16 0.2913 (Mariposa, 590 m) 1 -0.1108 0.0874 1.61 0.2051 (Moore, 830 m) 1 -0.0060 0.0898 0 0.9466 (Bear, 860 m) 1 -0.0013 0.0896 0 0.9888 (Snow, 950 m) 1 -0.0429 0.0889 0.23 0.6298 (Tenaya, 1210 m) 1 0 0 (Tuolumne, 1320 m) 1 0.0977 0.0919 1.13 0.2879 Site by Population 15 30.43 0.0105 (Levels not shown) Shape parameter 1 1.5715 0.0384 M Iewisii Site 3 4964.46 <0.0001 (Jamestown, 415 m) 1 -5.0646 0.2199 530.68 <0.0001 (Mather, 1400 m) 1 -2.0826 0.2182 91.11 <0.0001 (White Wolf, 2395 m) 0 0 0 (Timberline, 3010 m) 1 -0.6668 0.2529 6.95 0.0084 Population 5 3.06 0.6902 (Tuolumne, 1320 m) 1 0.1323 0.2852 0.22 0.6426 (Tamarack, 1910 m) 1 -0.3765 0.2527 2.22 0.1363 (Porcupine, 2400 m) 1 0.1583 0.2852 0.31 0.5789 (Tioga, 2580 m) 1 0.0476 0.2774 0.03 0.8636 (Snow, 2690 m) 1 -0.0463 0.2709 0.03 0.8643 (Warren, 2750 m) 0 0 0 Site by Population 15 16.05 0.3784 (Levels not shown) Shape mrameter 1 1.1263 0.0228 58 RESULTS Survivorship Table 5 gives the results of failure-time analyses. For both species, transplant site had a highly significant effect on survival time. All sites (Jamestown, 415 m, Mather, 1400 m, and Timberline, 3010 m) were significantly different from the reference site, White Wolf (2395 m), as indicated by regression coefficients different from zero. To examine differences among non—reference sites, I constructed Z-tests for comparisons of regression coefficients and found that all pairwise differences among non-reference sites were also significant, although for M cardinalis the difference between Mather (1400 m) and Timberline (3010 m) was only marginally significant after correcting for multiple comparisons (Table 6). Population did not affect survival time for either species. For M cardinalis, the population by site effect was significant, indicating that populations differ in their response to elevation. There was no population by site interaction for M Iewisii survivorship. For both species, the Weibull shape parameter was significantly greater than 1, indicating that the risk of mortality increased monotonically with time. Table 6. Pairwise differences of transplant site regression coefficients from accelerated failure-time analyses. After correcting for multiple comparisons, only Z-scores > 2.12 remain significant at the 0.05 level. M cardinalis M Iewisii Z P Z P Jamestown (415 m) vs. Mather (1400 m) 7.89 <0.0001 22.40 <0.0001 Jamestown (415 m) vs. Timberline (3010 m) 6.53 <0.0001 23.16 <0.0001 Mather (1400 m) vs. Timberline (3010 m) 1.93 0.0265 7.50 <0.0001 Mimulus cardinalis survival during the first year was highest at the 1400 m range border, intermediate at high elevations beyond the range, and lowest at the 415 m 59 ’63 A) M cardinalis B) M Iewisii t: :2 1 ‘ 1 d 77W --=—-—.~§7. fir;-: ‘3 E 'o ............. Q0 m 0 t 01 - o 1 4 EL —-— JA (415 m) ' _._ JA(415m) 3 ...... O ....... MA (1400 m) ....... Q ....... MA (1400 m) a ——+-- WW(2395 m) -—+—- WW(2395 m) *4 -~v-—~ TI(3010m) ——-47—-- TI (3010 m) 0.01 . . . . . 0.01 . . . . 0 200 400 600 8001000 0 200 400 600 8001000 Time (days) Time (days) Figure 7. Survivorship at each transplant site. A) M cardinalis. B) M Iewisii. Transplant site abbreviations as follows: JA = Jamestown, MA = Mather, WW = White Wolf, T1 = Timberline. range center (Figure 7a). There was an early decrease in survival during the first growing season at 415 m, whereas survival at 2395 and 3010 m was high during the first growing season and declined over the first winter. During subsequent years, survivorship remained highest at 1400 m and was reduced at 2395 and 3010 m. Examination of regression coefficient confidence limits for each site and population indicated that the M cardinalis population by site interaction arose because of differences in elevation response between the low elevation Mariposa Creek population (590 m) and the mid elevation Tenaya Creek population (1210 m; data not shown). At 1400 m, the Mariposa Creek population survived longer than the Tenaya Creek population, and the converse was true at 3010 m. Mimulus Iewisii survival was highest at 2395 and 3010 m and intermediate at 1400 m (Figure 7b). At 415 m, M Iewisii suffered high mortality during the first growing season. The few individuals surviving after one growing season at 415 m died over the winter, resulting in 100% mortality within one year. At 1400 m, M Iewisii 60 experienced pulses of mortality at the end of the second and third growing seasons. At high elevations, mortality rates were roughly constant and low. Growth For both species, site had a highly significant effect on growth, measured as log- transformed average annual stem length (Table 7). There were no significant population or population by site effects for M cardinalis growth, but both population and population by site effects significantly affected grth for M Iewisii. Sire, dam and all interactions involving sire or dam were non-significant for both species. Table 7. Linear mixed model analysis of variance summary for log-transformed average annual stem length. F -tests for fixed effects constructed by SAS MIXED procedure, with denominator degrees of freedom obtained from the Satterthwaite approximation and indicated in parentheses below each F -value. All random effects (sire, dam, and their interactions) were estimated to be zero or near-zero and were not significant. M cardinalis M Iewisii Fixed effects df F P (if F P Site 3 934.29 <0.0001 2 575.82 <0.0001 (1284) (1264) Pop 5 1.17 0.3197 5 5.85 0.0001 (1284) (70.4) Site*Pop 15 0.45 0.9629 10 2.58 0.0043 (1284) (1263) Mimulus cardinalis growth was greatest at 415 m, intermediate at 1400 m, and greatly reduced at higher elevations (Figure 8a). Growth of M Iewisii peaked at 1400 and 2395 m and was reduced at 3010 (Figure 8b). The difference in grth between the 1400 m range margin and the 2395 m range center was not statistically significant in Tukey- Kramer adjusted post-hoe contrasts. High mortality resulted in small sample 61 W O b.) O fit) A) M cardinalis B) M Iewisii a 5 a 25 a 26.6 ‘ 60 ' .v; E E 20 - I "7’ g g 340 15 - '{f‘j o E” 10 ‘ g0 20 2 5 - < it. 0 ‘ 0 ' JA MA WW TL JA MA WW TL 415 1400 2395 3010 415 1400 2395 3010 Transplant site (m) Transplant site (m) Figure 8. Species average annual stem length + SE at each transplant site (mean values given within each bar). A) M cardinalis B) M Iewisii. Site means sharing the same letter are not significantly different. Note that species are graphed on different scales. Transplant site abbreviations as in Figure 7. 80 60 A) M cardinalis B) M Iewisii _ Ma, 590 50 4 _ Tu, 1320 60 1 :1 M0, 830 1:1 Ta, I910 _ Be, 860 40 4 _ Po,2400 1:: Sn, 950 :1 Ti, 2580 — Te,1210 30 a — Sn, 2690 _ Tu, 1320 _ Wa, 2750 Average annual stem length (cm per year) A O 20 4 20 - 10 ~ 0 - ___~ 0 - _ _ a a b c g 5.08?ng d 0 JA MA WW TI JA MA WW 415 1400 2395 3010 415 1400 2395 3010 Transplant site (m) Transplant site (m) Figure 9. Population average annual stem length + SE versus transplant site. A) M cardinalis. B) M Iewisii. Populations are arrayed in order of increasing elevation of origin. Population means sharing the same letter are not significantly different. Note that species are graphed on different scales. Transplant site abbreviations as in Figure 7. Population abbreviations as follows: Ma = Mariposa Ck., Mo = Moore Ck., Be = Bear Ck., Sn = Snow Ck., Te = Tenaya Ck., Tu = S. Fork Tuolumne R., Ta = Tamarack Ck., P0 = Porcupine Ck., Ti = Tioga Rd., Sn = Snow Ck., Wa = Warren Fork Lee Vining R. 62 size for M Iewisii at 415 m (N=2). The M Iewisii population effect was due to the difference between the Warren Fork population (2750 m) and all other populations except for the South Fork population (1320 m; Figure 9). The Warren Fork population reached a smaller size than other M Iewisii populations regardless of site. The M Iewisii population by site interaction indicated that populations differed in their grth response to elevation. This difference was driven by the greater increase in growth at 2395 m versus 3010 m for two mid elevation populations (South Fork, 1320 m, and Tamarack Creek, 1900 m) relative to two high elevation populations (Snow Creek, 2690 m, and Warren Fork, 2750 m). Fitness Transplant site strongly affected average annual fitness of both species (Table 8). Population of origin and the interaction between population and transplant site had marginally significant effects on M cardinalis fitness and highly significant effects on M Iewisii fitness. Sire and dam components of variance were not significant for either species. For M cardinalis, the sire by site interaction was significant, and for M Iewisii, the darn by site interaction was significant. The existence of sire or dam by site interactions in these species is consistent with the presence of genetic variation for fitness across elevations. However, examination of sire and darn means revealed high variance and large heteroskedasticity of variance across sites (data not shown), making it more likely that the significance of these interaction effects is an artifact of variance (Juenger and Bergelson 2000). 63 Table 8. Generalized linear mixed model analyses of average annual fitness. F-tests for fixed effects constructed by SAS MIXED procedure, with denominator degrees of freedom obtained from the Satterthwaite approximation and indicated in parentheses below each F-value. Z-tests for random effects constructed by ‘covtest’ option. M cardinalis M Iewisii Fixed effects df F P F P Site 3 474.48 <0.0001 78.21 <0.0001 (70) (143) Pop 5 2.90 0.0722 3.98 0.0036 (41.9) (42) Site*Pop 15 1.82 0.0635 3.76 <0.0001 (69.9) (142) Random effects Estimate P Estimate P Sire(Pop) 0.005 0.2231 0 Dam(Pop) 0 0.001 1 0.1407 Sire*Dam(Pop) 0 0 Sire*Site(Pop) 0.055 <0.0001 0 Dam*Site(Pop) 0.015 0.0535 0.006 0.0016 Sire*Dam*Site(Pop) 0 0.000 0.4750 Error 0.312 0.098 Mimulus cardinalis fitness was highest at the 415 m range center, reduced at the 1400 m range border, and zero or near-zero at high elevations beyond its present range (Figure 10a). The population from the lowest elevation of origin, Mariposa Creek (590 m), was more fit than a population from middle elevation, Tenaya Creek (1210 m; Figure 11a). The population by site interaction was driven by populations differing in the degree of decrease in fitness from 1400 m to 2395 m. The Bear Creek population (860 m) displayed a greater decrease in fitness from 1400 to 2395 m than did the Tenaya Creek population (1210 m; Figure 11a). Mimulus Iewisii fitness was highest at the 2395 m range center, intermediate at the 1400 m and 3010 range borders, and lowest at 415 m, beyond its present range (Figure 10b). The Tioga Road population (2580 m) was more fit than the population 64 O\ 0.6 A) M cardinalis B) M Iewisii * a 0.5 ~ .35. 0.4 - v—I :— _l C N 1 0.3 - 0.2 - c c 0.1 - a m , 0.1 0 0.0 0 JA MA WW TI JA MA WW T1 415 1400 2395 3010 415 1400 2395 3010 Transplant site (m) Transplant site (m) Average annual fitness (flower number per year) 00 Figure 10. Species average annual fitness (in units of flowers per year) + SE versus transplant site (mean values given within each bar). A) M cardinalis B) M Iewisii. Site means sharing the same letter are not significantly different. Note that species are graphed on different scales. Transplant site abbreviations as in Figure 7. 25 1.0 A) M cardinalis B) M Iewisii fiflzo - — Ma, 590 0.3 -_ Tu,1320 3 § M0, 830 a: Ta, 1910 Te :15 4 — Be, 860 0;, - — Po,2400 E ‘3’. :1 Sn, 950 1:1 Ti,2580 3 510 . — Te,1210 Q4 - — Sn, 2690 5"?) — Tu, 1320 0 C3 2 v 5 ‘ E 0 - _ _ __ __ ”agagggg'gjv r f JA MA WW TI 415 1400 2395 3010 415 1400 2395 3010 Transplant site (m) Transplant site (m) Figure 11. Population average annual fitness + SE versus transplant site. A) M cardinalis. B) M Iewisii. Populations are arrayed in order of increasing elevation of origin. Population means sharing the same letter are not significantly different. Note that species are graphed on different scales. Abbreviations as in Figure 9. 65 from the highest elevation, Warren Fork (3010 m), across all sites (Figure 11b). The population by site effect indicated that populations differ in their reaction norms for fitness versus elevation. This interaction was the result of populations differing in the degree of increase in fitness at 2395 m, relative to the uniformly low fitness at other sites. The South Fork (1320 m), Tamarack Creek (1920 m) and Tioga Road (2580 m) populations showed a large increase in fitness at 2395 m, whereas the Porcupine Creek (2400 m), Snow Creek (2690 m) and Warren Fork (2750 m) populations did not show a statistically significant increase in fitness at 2395 m (Figure 11b). To determine whether populations are adapted to their position within the elevation range, I also examined the rank correlation between average fitness and the difference in elevation between transplant site and population origin. If populations are adapted to position within the elevation range, then the correlation between fitness and the difference between origin and transplant elevations should be negative, indicating that fitness declines as the transplant environment becomes more different from the native environment. No correlations were statistically significant, suggesting that fitness variation among populations is not caused by differences in elevation of origin (Table 9). Table 9. Rank correlation between population average annual fitness and transplant elevation — population origin elevation . M cardinalis M Iewisii Site r Prob > |r| r Prob > |r| 415 -0.0212 0.9024 0.0289 0.8669 1400 0.0683 0.6924 -0.0220 0.8988 2395 -0.0801 0.6425 0.0361 0.8344 3010 0.0084 0.9610 0.1076 0.5324 66 DISCUSSION Geographic Variation in Fitness The results of this reciprocal transplant experiment support the hypothesis that species are most fit at their range center and become increasingly maladapted as the distance from the range center increases. Both species exhibited the greatest average fitness at elevations central within their range (415 m for M cardinalis, 2395 m for M Iewisii) and reduced fitness at elevations at the range margin (1400 m for both species, 3010 m for M Iewisii). F urtherrnore, both species exhibited zero or near-zero fitness when transplanted beyond their present elevation range limits (to higher elevations, 2395 and 3010 m, for M cardinalis, or to lower elevation, 415 m, for M Iewisii). However, the underlying causes of this fitness variation differed between the species. For M cardinalis, first-year survival was relatively high across all elevations, but growth and fecundity were higher at the low elevation range center than at higher elevations. At higher elevations, few M cardinalis individuals were able to reach reproductive maturity. Individuals that flowered at 2395 m did so in September, after most M Iewisii stopped flowering, and did not mature seeds before senescence. By contrast, M Iewisii confronted a strong survival barrier at its lower elevation range limit. Mortality during the first growing season at 415 m was rapid; most individuals died within one month of planting and all were dead within one year. Because experimental planting was timed to match the phenology of natural populations, transplanted seedlings were exposed to the climate they would have encountered if naturally dispersed to low elevation. A large preliminary study conducted at 415 m in June 2000 produced nearly identical results (M cardinalis survival: 85.8% after four 67 months, 76.3% after 10 months, N=962; M Iewisii survival: 6.2% after four months, 0% after 10 months, N=953), indicating that observed patterns of mortality are not exaggerated by unusually harsh conditions in 2001. These findings are largely congruent with the patterns of variation in performance across elevation in these species described by Hiesey et al. (1971) in their landmark reciprocal transplant study of M cardinalis and M Iewisii. They demonstrated that low survival and reproductive capacity of M cardinalis at high elevation and low survival and growth of M Iewisii in a coastal climate. However, in their study, M cardinalis displayed the highest survivorship in the low elevation Stanford transplant garden (30 m), whereas I observed highest first-year survivorship in the mid elevation Mather garden. This difference highlights the important difference between the low elevation maritime environment and the low elevation foothills environment. A second difference between the present findings and the previous study is the relatively poor performance of M Iewisii that I observed at Timberline, where Hiesey et al. (1971) found that M Iewisii achieved its highest performance. This difference is likely due to several factors, including the addition of White Wolf as an intermediate transplant site between Mather and Timberline, exclusion of populations from the northern race of M Iewisii, and use of seedlings rather than vegetatively propagated clones. The use of seedlings provided important information about the performance of early life history stages, which may experience strong selection and be critical for population establishment (Lee et al. 2003; Zacherl et al. 2003). It is also important to note that none of the transplant sites used by Hiesey et al. (1971) were central within the elevation range of M Iewisii. 68 Several other experiments have demonstrated reduced growth, delayed phenology, and, as a result, reduced fecundity of plant species transplanted beyond their northern or high elevation range margins (Prince 1976; Davison 1977; Woodward 1990; Asselin et al. 2003). Analogous patterns of delayed development have also been reported for aphids (Gilbert 1980) and butterflies (Crozier 2004) transplanted beyond their latitudinal range limits. In these examples, fitness reductions generally are not due to a single environmental event such as a frost or to a single vulnerable life history stage, but rather result from the gradual accumulation and cascading effects of fitness reductions at many stages. In contrast to expectations for northern or upland range limits, it is generally assumed that climate becomes more permissive for most organisms and that biotic interactions become relatively more important in setting southern or lowland distribution limits (MacArthur 1972; Woodward 1975; Sievert and Keith 1985; Hersteinsson and Macdonald 1992; Richter et al. 1997; Scheidel et al. 2003; Cleavitt 2004). Few studies of southern or lowland distributions limits find severe abiotic limitation as I have documented for M Iewisii at low elevations. Many plants showed signs of heat stress such as leaf scorching and reduced leaf size, and subsequent grth chamber studies have demonstrated strikingly similar patterns of mortality when M Iewisii are grown under the high temperatures characteristic of low elevation (Chapter 3). Population Variation in Fitness Although population and population by site effects were frequently statistically significant, they were of much smaller magnitude than site effects. I detected 69 differences among M cardinalis populations, but not M Iewisii populations, in survivorship at different elevations. For M cardinalis, the low elevation Mariposa Creek population (590 m) survived longer at middle elevation than the mid-elevation Tenaya Creek population (1210 m), but at high elevation the Tenaya Creek population had survived longer than the Mariposa Creek population. The direction of reversal in survivorship is consistent with adaptation of the range margin Tenaya Creek population to higher elevations, but it is not entirely consistent with adaptation to position within the range because of the poor relative performance of the mid-elevation population at middle elevation. No other differences among populations were significant, indicating that differentiation among populations for survivorship is low. I detected differences among M Iewisii populations but not M cardinalis populations in average annual stem length. Local adaptation of growth traits may take two possible forms. First, populations could exhibit genetically based clinal differences in growth in which populations originating from higher elevations display reduced grth rates or short stature across all environments (Clausen et a1. 1940). Alternatively, populations could show decreasing growth with increasing distance from population origin. I find some slight evidence that the former scenario is true for M Iewisii. Differences among M Iewisii populations were consistent with a trend for genetically based clinal differences in average annual stem length, where populations from the mid elevation range margin reached larger size at 2395 m and the population from the highest elevation of origin was smallest at both 2395 and 1400 m. Hiesey et al. (1971) also found some evidence for genetically based clinal growth differences among M Iewisii populations. However, because in their study populations were collected from 70 throughout the geographic ranges of both species, the wide latitudinal and longitudinal distances that separated most populations from the transplant sites are not easily separated from the effects of adaptation to environmental variables that vary with elevation. For both species, I detected variation among populations in reaction norms for average annual fitness versus transplant site, but these differences were not consistent with the hypothesis that populations are adapted to their elevation of origin. For example, at Mather (1400 m), the nearby South Fork populations of both species (1320 m) were not more fit than the distant M cardinalis Mariposa Creek (590 m) or M Iewisii Warren Fork (2750 m) populations. The marginally significant M cardinalis population by site interaction resulted solely from two populations differing in the degree of decrease in fitness from 1400 m to 2395. The significant M Iewisii population by site interaction arose because three populations (South Fork, 1320 m, Tamarack Creek, 1920 m, and Tioga Road seep, 2580 m) displayed significantly increased fitness at 2395 m versus other elevations and three did not (Porcupine Creek, 2400 m, Snow Creek, 2690 m, and Warren Fork, 2750 m). Reaction norms for fitness never crossed, but instead differed in the slope of decrease from the range center to range margins, suggesting that populations do not exhibit symmetrical “home” elevation advantages. This conclusion is supported by the lack of significant correlations between population mean fitness and the difference in elevation between p0pulation origin and transplant site. Gene Flow and Selection 71 Range limits arise where populations are no longer able to adapt sufficiently to local environmental conditions. Low fitness of both species at their range margin suggests that adaptation to the marginal environment is hindered. Likewise, weak differentiation among populations within each species indicates that populations from the range margin have been unable to adapt to environmental conditions at the range boundary. The lack of adaptation to elevation of origin that I observe is striking given the number of documented examples of adaptive differentiation both among populations at geographic scales (e.g., Clausen et al. 1940; Grant 1963) and within populations at extremely local spatial scales (e.g., Bradshaw 1960; Schemske 1984). Many species display ecotypic variation along altitudinal gradients (Clausen et al. 1940; Oleksyn et al. 1998; Jonas and Geber 1999). The populations used in this experiment were sampled along an elevation gradient that imposes variation in several important abiotic environmental variables, including length of growing season and temperature. Species may not be able to adapt to environmental conditions at the range margin if they lack appropriate genetic variation upon which selection can act or if differential natural selection is weak relative to the homogenizing effects of gene flow (Mayr 1963; Kirkpatrick and Barton 1997). The interplay of gene flow and selection along environmental gradients or between discrete environments is important to several models of range or niche evolution (Holt and Gaines 1992; Kawecki 1995; Kirkpatrick and Barton 1997; Gomulkiewicz et al. 1999; Holt 2003). For example, Kirkpatrick and Barton (1997) modeled the evolution of a quantitative character determining fitness across a one- 72 dimensional environmental gradient. The character evolved under stabilizing selection toward an optimum phenotype that varied with the environmental gradient. Population density in their model depended on dispersal, density-dependent population regulation and the degree of mismatch between the optimum and population mean phenotypes. Stable range limits arose when gene flow imposed a strong constraint on local adaptation, as when dispersal was high or the environmental gradient was steep. Although the focus of the Kirkpatrick and Barton model was on the swamping effects of gene flow, it also modeled adaptive trade-offs between environments because no single phenotype was optimal across the entire environmental gradient. Models of niche evolution explicitly consider the role of trade-offs between habitats in limiting species distributions, finding that selection to improve adaptation to environments outside of the niche may be weak due to the demographic asymmetry between habitats within versus outside of the niche (Kawecki 1995; Holt 1996; Gomulkiewicz et al. 1999). In a recent model of range evolution, Holt (2003) explicitly modeled the feedback between the evolution of dispersal and the evolution of habitat specialization (i.e., trade-offs) in a two-habitat model where neither habitat was initially outside of the niche. In this model the evolutionary dynamics of the geographic range depended on the shape of adaptive trade-offs between habitats and the initial habitat distribution of the population. For instance, a species initially specialized to one habitat may evolve habitat generalization if mutations that increase adaptation to a new habitat have little cost to fitness within the present habitat. Conversely, if a linear and symmetrical trade-off in fitness between two habitats exists, evolution will favor increased specialization to whichever habitat the species initially resides in. These models highlight the need to 73 understand the relative roles of dispersal, adaptive trade-offs and demographic asymmetries between habitats in range evolution. Further work is necessary to understand how these components interact to determine the elevation range limits of Mimulus cardinalis and M Iewisii. Dispersal .— Elevation distributions offer a tractable experimental analog to latitudinal distributions at larger spatial scales, because both arise along continuous environmental gradients and encompass multiple populations. The environmental gradient from the center to the edge of elevation and latitude ranges is also similar, with temperature and length of growing season decreasing to the north and at higher elevations, although the rate of change in environmental parameters across space is greater for altitudinal than for latitudinal gradients. Indeed, a change of 100-200 m in elevation is roughly equivalent to a change of 1° in latitude (Criddle et al. 1994; Flebbe 1994). Due to the steepness of the enviromnental gradient across elevation, for a given dispersal distance, individuals encounter a more different environment than if dispersing across latitude, making it more likely that marginal populations may be swamped by centrally adapted phenotypes at altitudinal than at latitudinal range boundaries (Kirkpatrick and Barton 1997). Little is known about mechanisms of dispersal of M cardinalis and M Iewisii seeds. Because both species occur in riparian habitats, it is possible that seed dispersal via downstream currents provides a mechanism for primarily unidirectional long- distance dispersal among populations, setting up an interesting dichotomy between M cardinalis and M Iewisii at their shared mid elevation range boundary. A net flux of migrants downstream would imply that the M. Iewisii mid elevation range limit may be 74 subject to swamping gene flow from high elevation central populations, but that the M cardinalis mid elevation range limit is not. However, gene flow via pollen may show the opposite pattern due to the greater flight distance of hummingbirds, the primary pollinator of M. cardinalis, compared to bumblebees, the primary pollinator of M Iewisii. Estimations of F 5. among populations of each species are in progress to begin to identify patterns of gene flow among central and marginal populations of each species. Adaptive trade-ofls.— Because central and marginal populations of each species display few adaptive differences versus elevation, interspecific comparisons are necessary to understand adaptive trade-offs across the elevation gradient. Since their recent common ancestor, M cardinalis and M Iewisii have evolved differences that restrict their distributions to different areas of the complex environmental gradient associated with elevation. Specialization to different elevation ranges suggests that different phenotypes are necessary for fitness at low versus high elevations. Estimation of the strength and direction of selection on phenotypic traits across the elevation gradient, in combination with genetic mapping of quantitative trait loci, will identify traits under selection at high versus low elevation and the underlying genetic architecture of those traits (Angert, Bradshaw, and Schemske, unpub. data). Experimental evolution of segregating hybrid populations at low and high elevation will also illuminate whether there are fitness costs of specialization to low versus high elevation (Angert, Bradshaw, and Schemske, unpub. data). Together, these studies will help elucidate mechanisms of adaptive trade-offs between low and high elevation environments. In conjunction with estimates of gene flow between central and marginal 75 populations, we can hope to understand what causes and constrains adaptation to different elevation ranges. 76 CHAPTER 3 Growth and leaf physiology of monkeyflowers (Mimulus cardinalis and M Iewisii) with different elevation ranges Abstract—Every species is limited both geographically and ecologically to a subset of available habitats, yet for many species the causes of distribution limits are unknown. Temperature is thought to be one of the primary determinants of species distributions along latitudinal and altitudinal gradients. This study examined leaf physiology and plant performance under contrasting temperature regimes of sister species of monkeyflower, Mimulus cardinalis and M Iewisii (Phrymaceae), that differ in elevation distribution to test the hypothesis that temperature-dependent differences in growth are an important determinant of differences in fitness versus elevation. Each species attained greatest aboveground biomass, net photosynthetic rate, and effective quantum yield of photosystem II when grown under temperatures characteristic of the altitudinal range center. Although both species exhibited greater stem length, stomatal conductance, and intercellular CO; concentration in hot than in cold temperatures, these traits showed much greater reductions under cold temperature for M cardinalis (native to low elevation) than for M Iewisii (native to high elevation). Survival of M Iewisii was also sensitive to temperature, showing a striking decrease in hot temperatures. Within each temperature regime, the species native to that temperature displayed greatest growth and leaf physiological capacity. Populations from the elevation range center and range margin of each species did not differ in most grth or leaf physiological responses to temperature. This study provides evidence that M cardinalis 77 and M Iewisii differ in survival, growth, and leaf physiology under temperature regimes characterizing their contrasting low and high elevation range centers, and suggests that the species’ elevation range limits may arise, in part, due to metabolic limitations on growth that ultimately decrease survival and limit reproduction. Key words: range boundary, distribution limit, elevation, temperature, photosynthesis No species occupies an unlimited area. Rather, every species is limited both geographically and ecologically to a subset of available habitats. Understanding the patterns and processes governing the distribution of species is a central goal of ecology, yet for many species the causes of distribution limits are unknown. Identifying the causal mechanisms of distribution limits is challenging because environmental variables are often spatially correlated and dissecting organismal responses to even a single environmental variable is a complex task. However, temperature is thought to be one of the primary determinants of species distributions along latitudinal and altitudinal gradients. Evidence for the role of temperature in distribution limits comes from a diverse array of studies, including correlations between isotherms and distribution boundaries (e.g., McNab 1973; Grace 1987; Root 1988b), temperature tolerance and latitudinal or altitudinal distribution (e. g., Loik and Nobel 1993; Cunningham and Read 2002; Kimura 2004), extreme temperature events and periods of reproductive failure or high mortality at range boundaries (e.g., Silberbauer- Gottsberger et al. 1977; Jarvinen and Vaisanen 1984; Olmsted et al. 1993; Mehlman 1997), and studies of latitudinal and altitudinal changes in response to both historic and recent global warming trends (e. g., Huntley 1991; Parmesan et al. 1999; Hughes 2000; 78 Thomas et al. 2001). Further, temperature exerts a ubiquitous influence on many important cellular properties such as the rate of enzymatic reactions, protein conformations and membrane stability. Temperature may influence species distributions in a multitude of ways, from imposing direct lethal limits to regulating processes of growth, development and reproduction (Cossins and Bowler 1987; Orfanidis 1993; Molenaar and Breeman 1994; Sewell and Young 1999). Study of the sensitivity of metabolic processes to temperature can elucidate the mechanisms underlying limitation at distribution boundaries (Heller and Gates 1971; McNab 1973; Criddle et al. 1994; Anthony and Connolly 2004). For plants, photosynthesis is a primary metabolic process and is the source of energy and substrates for all other biosyntheses. Photosynthesis often exhibits a temperature optimum, deviations from which cause photosynthetic activity to decrease (Larcher 1995; Battaglia et a1. 1996). Populations or species from contrasting temperature habitats often exhibit differences in photosynthetic optima and acclimation ability in response to temperature (Billings et al. 1971; Berry and Bjorkman 1980; Amtz and Delph 2001; Cunningham and Read 2003). Other gas exchange parameters are also sensitive to temperature. Without stomatal regulation, transpiration rises with rising temperature. However, extremes of temperature often elicit stomatal closure, which may decrease stomatal conductance and limit the availability of C02 for photosynthesis (Larcher 1995). Long-term acclimation to temperature may alter stomatal conductance as a result of changes in stomatal density or aperture (Ferris et al. 1996). Measurement of instantaneous leaf gas exchange parameters such as net photosynthetic rate and stomatal conductance offer a way to 79 detect functional limitations on plant metabolism imposed by environmental factors (Llorens et a1. 2004). Chlorophyll a fluorescence provides another non-destructive means to assess the functioning of the photosynthetic system. Light energy absorbed by a leaf can be used for photochemical reactions, dissipated as heat energy, or re-emitted as fluorescent light (Bolhar-Nordenkampf and Oquist 1993). The measured fluorescence signal from a leaf is determined by the rate constants of these competing reactions and the fraction of open reaction centers available for photochemistry and comes primarily from chlorophyll a of photosystem 11 (PS 11; Krause and Weis 1984). Measurement of chlorophyll fluorescence of light-adapted leaves can determine the fraction of absorbed light energy used in electron transport, or the effective quantum yield of photosystem 11 ((1112511). Thylakoid membranes are especially sensitive to heat and chilling, so disturbance of photosynthesis, particularly in PS 11, is a first sign of temperature stress (Berry and Bjorkman 1980; Bolhar-Nordenkampf and Oquist 1993). Even when thylakoid membranes remain intact, temperature stress may decrease fluorescence yield due to down-regulation of PS 11 activity and increases in non-photochemical quenching resulting from the inhibition of carbon metabolism (Krause and Weis 1991; Owens 1994; Schreiber et al. 1994; Haldimann and Feller 2004). Thus, measurement of chlorophyll fluorescence can detect early stages of both low and high temperature stress. This study examines leaf physiology and plant performance under contrasting temperature regimes of sister species of monkeyflower, Mimulus cardinalis and M Iewisii (Phrymaceae), that differ in elevation distribution. Reciprocal transplants 80 demonstrate that each species has high growth, survival and reproduction at its elevation range center and lower growth, survival and reproduction at its elevation range boundary and at elevations beyond its present elevation range (Chapter 2). Here I test the hypothesis that temperature is an important determinant of these differences in plant performance using temperature regimes measured in the field to simulate natural low and high elevation environments during the growing season. To examine adaptive differentiation among populations, populations from the elevation range center and range margin of each species were used as source material for the experiment. Specifically, this study asks 1) do M cardinalis and M Iewisii differ in performance under temperature regimes characterizing their contrasting low and high elevation range centers? and 2) Do differences in leaf physiological traits underlie differences in performance under contrasting temperature regimes? MATERIALS AND METHODS Study System Mimulus cardinalis and M Iewisii (Phrymaceae) are rhizomatous perennial herbs that grow along seeps and stream banks in western North America. Both species are self-compatible and animal pollinated (Hiesey et al. 1971; Schemske and Bradshaw 1999). Mimulus cardinalis occurs from southern Oregon to northern Baja California, Mexico and from the coast of California inland to Arizona and Nevada. Mimulus Iewisii is composed of two races, a northern form occurring from southern coastal Alaska to southern Oregon and eastward to the Rocky Mountains, and a southern form, occurring primarily in the Sierra Nevada Mountains of California (Hiesey et a1. 1971; Hickman 1993; Beardsley et al. 2003). The two races are partially incompatible, and recent 81 phylogenetic analysis suggests that the two races are sister to one another and together are sister to M cardinalis (Beardsley et al. 2003). Here I study only the Sierran form of M Iewisii. Mimulus cardinalis and M Iewisii segregate by elevation, with M cardinalis occurring from sea level to 2400 m and M Iewisii occurring from 1200 m to 3100 m in California (Hickman 1993). In the Yosemite National Park region where this research was conducted, the species co-occur on larger watercourses between 1200 and 1500 m elevation (A. Angert, unpub. data). Although the published Californian distributions of M cardinalis and M Iewisii extend to 2400 and 3100 m, respectively, repeated attempts to locate extant populations at these upper limits in the Yosemite region were unsuccessful. Experimental gardens planted at 415, 1400, 2395 and 3010 m on the western slope of the Sierra Nevada Mountains demonstrate that each species is most fit at its elevation range center, (415 m for M cardinalis, 2395 m for M Iewisii), less fit at the mid-elevation range boundary, and unable to both survive and reproduce when transplanted to elevations beyond its current range (Chapter 2). For M Iewisii, reduced fitness at low elevations results primarily from high juvenile mortality within the first growing season. For M cardinalis, reduced fitness at high elevations is due primarily to limited growth and reproduction (Chapter 2). Genetic Material: Population Collection and Crossing Design Seeds from eight plants in each of four populations per species were collected in September 1999 along an elevation gradient from 590 m to 2750 In between 37.49 and 37.95 ° N latitude (Appendix A). For each species, the chosen populations represent two locations from central within the range (low elevation for M cardinalis, high elevation 82 for M Iewisii) and two locations from the range margin (mid elevation for both species). One plant from each field-collected family was grown to flowering in the University of Washington greenhouse under standard greenhouse conditions. The eight plants from each population were crossed with one another so that each plant served as sire or dam once with no self- or reciprocal pollinations, generating four independent full-sib families. Pollinations were performed by collecting all of the pollen from one flower with a flat toothpick and fully saturating the stigma of one flower. Seeds from four pollinations per full-sib family were pooled. These crosses generated outcrossed seeds from each population in a uniform environment to be used for controlled environment studies. Chamber Conditions Two incubators (Model I-36LL, Percival Scientific, Perry, IA, USA) were programmed to simulate low and high elevation temperature regimes for 60 days. To determine representative low and high elevation temperatures regimes during the growing season, data loggers (Hobo Pro Temp/Extemal Temp, Onset Computer Corp., Bourne, MA, USA) recorded temperatures at low (415 m, near Jamestown, California) and high (2395 m, at the White Wolf Ranger Station in Yosemite National Park, California) elevation sites during June - September 2002. These sites have been used for reciprocal transplant gardens (Chapter 2) and are concordant with the range center of M cardinalis and M Iewisii, respectively. Two data loggers at each site recorded air temperature every half hour. Loggers were mounted at plant height and shielded from direct sunlight with reflective covers. 83 Table 10. July 2002 mean temperatures recorded in reciprocal transplant gardens at 415 and 2395 m. Temperature (°C) Elev.(m) Ave. daily Max. daily Days >40 Ave. daily Min. daily Days < 0 max. max. min. min. 415 34.48 41.67 4 14.80 11.67 0 2395 22.78 27.91 0 4.28 -1.97 2 Incubator temperature programs were set to reflect July average daily maximums and minimums at each elevation, with occasional temperature spikes or dips occurring at natural frequency (Table 10). July temperatures were used because plant growth is at its peak at both low and high elevation during this time. The cold, high elevation chamber was set for a 23 °C daytime maximum and 4 °C nighttime minimum, with one 0 °C freeze on night 15 and a second -2 °C freeze on night 36. Although few plants showed visible signs of tissue injury after exposure to 0° C, many plants were injured by the second, -2° C freeze. To quantify tissue damage, I estimated the percentage of total leaf tissue damaged on each plant. The hot, low elevation chamber was set for a 35 °C daytime maximum and 15 °C nighttime minimum, with 42 °C daytime maximums on days 18, 30, and 51. Daily maximum and minimum temperatures (including extremes) were held for four hours each with gradual ramps between maximum and minimum temperatures. Incubators were programmed for 14/10 hour day/night cycles with the maximum possible light output, 200 umol photons rn'2 s' 1 during the daytime period. In natural environments M cardinalis and M Iewisii grow in a range of light conditions from full sun on open gravel bars to full shade along riparian corridors (A. Angert, pers. obs.). 84 Four replicates of each full-sib family were sown in the Michigan State University greenhouse in January 2003. Five weeks after sowing, seedlings were transferred to either the hot or the cold incubator, for a total of 64 plants per temperature treatment (2 species x 4 populations / species x 4 families / population x 2 replicates per family). Seedlings were placed in random order within wire frames, and wire frames were placed in trays for sub-irrigation within the incubator. Frames were rotated several times per week to minimize position effects. Plants remained in each incubator for 60 days. Leaf Physiological Trait Measurements Simultaneous gas exchange and chlorophyll fluorescence measurements were performed following the last extreme temperature event for each treatment (day 53 hot, day 37 cold) with a portable open-flow gas exchange system equipped with leaf chamber fluorometer and C02 mixer (Li-Cor 6400, Li-Cor, Inc., Lincoln, NE, USA). The difference in time period preceding gas exchange measurements reflects natural differences in growing season length at low and high elevations. However, measurements made after the second extreme heat spike did not produce qualitatively different results, demonstrating that the patterns presented here are not unduly influenced by the length of exposure to low versus high temperatures. Measurements were made at midday during the 4-hour daily temperature maximum so that chamber temperature settings were not ramping throughout the course of the measurements. Because of sub-irrigation, plants were not water limited and gas exchange rates remained high at midday. This is realistic because M cardinalis and M Iewisii normally inhabit stream banks or permanent seeps. 85 The youngest fully-expanded leaf (second or third node) was enclosed within the leaf chamber. Instantaneous net photosynthetic rate (A, umol CO2 m'2 s"), stomatal conductance to water vapor (gs, mol H2O In2 S"), and the ratio of intercellular to ambient CO2 concentration (Ci/Ca) were determined at the light intensity in which 2 s", a reference CO2 concentration of 400 pmol leaves developed, 200 umol photons m' mol", a flow rate of 500 umol s'l’ and block temperatures of 35 ° C (hot chamber) or 23° C (cold chamber). Stomatal conductance is an indicator of the degree of stomatal openness, which determines leaf loss of water and gain of carbon dioxide, and the ratio of intercellular to ambient CO2 can indicate the degree to which stomatal closure limits the availability of CO2 for photosynthesis. Calculations of stomatal conductance assumed a 0.5 ratio of conductances on the upper versus lower side of each leaf. Before statistical analysis, stomatal conductance at high temperatures were reduced by 2% per °C above 23 °C to normalize for decreased water viscosity with increased temperature (Tyree et al. 1995; Sack et a1. 2002). Vapor pressure deficit and relative humidity within the leaf chamber were not controlled. Leaf temperature (°C) was measured with a fine wire thermocouple on the underside of each leaf. Steady-state fluorescence (F s) and maximal light-adapted fluorescence during a saturating flash of light (Fm’) were also measured simultaneously with gas exchange. These fluorescence parameters were used to calculate the effective quantum yield of photosystem 11 (0pm = (Fm’ — Fs)/Fm’), or the fraction of absorbed photons that a light-adapted leaf uses for photochemical reactions. Measurement of Plant Performance To quantify overall plant performance in each temperature environment, I measured final survival and growth. Traits were measured on day 60, at which time 86 plants were harvested to measure total stem length, number of nodes per stem, and aboveground biomass. Stern length and node number were highly correlated (M cardinalis: Pearson’s r=0.95, P<0.0001; M Iewisii: r=0.79, P<0.0001), whereas stem length and biomass were less so (M cardinalis: r=0.78, P<0.0001; M Iewisii: r=0.21, P=0.11), thus I present only stem length and biomass data. Statistical Analysis I performed mixed model analysis of variance (ANOVA) for both temperatures and species combined to model variation in each trait (A, (bpsu, normalized gs, Ci/Ca, aboveground biomass, and height) with respect to growth temperature, species, elevation of origin nested within species, population of origin nested within elevation, family nested within population, and all interactions. 1 also performed mixed model ANOVA within each temperature treatment to examine the effects of species, population, elevation of origin, and family on leaf temperature. Stomatal conductance and aboveground biomass were log-transformed to meet ANOVA assumptions. Temperature, species and elevation of origin were considered as fixed effects, whereas population and family were considered as random effects. To evaluate the significance of fixed effects, I used Type III estimable functions, which tolerate unbalanced samples, with denominator degrees of freedom obtained by Satterthwaite’s approximation. Intraspecific differences between temperatures and interspecific differences within each temperature were evaluated by independent contrasts with a single degree of freedom. Likelihood-ratio tests (comparing each reduced model to the full model including all effects) were used to evaluate the significance of all random effects. 87 To examine variation in post-freeze tissue damage, I performed mixed model analysis of variance as described above, with the following exceptions. Differences in post-freeze tissue damage were examined within the cold temperature regime only, thus the model included only species, elevation, population, and family effects. For this model I also included position within the incubator as a covariate to account for an unexpected temperature gradient from the front to the back of the chamber during the freeze. I did not model variation in survival with respect to growth temperature because no M cardinalis died in the hot temperature treatment, causing model convergence problems. All analyses were implemented with PROC MIXED in SAS, version 8.2 (SAS Institute, Inc., Cary, NC, USA). RESULTS Leaf Physiological Traits Table 11 gives the results of mixed model analysis of variance of instantaneous net photosynthesis, effective quantum yield, stomatal conductance (log-transformed and normalized to correct for temperature-induced changes in water viscosity), and Ci/Ca. The main effect of temperature affected stomatal conductance and Ci/Ca but not photosynthetic rate or effective quantum yield. The main effect of species only marginally affected effective quantum yield. However, species by temperature interactions affected all four parameters, indicating that the species differ in their leaf physiological response to temperature. Elevation of origin did not affect any leaf physiological trait, and the elevation by temperature interaction affected M cardinalis stomatal conductance and Ci/Ca only, indicating that differentiation in leaf physiological 88 Table 11. Linear mixed model analysis of variance summary for four leaf physiological traits: instantaneous net photosynthetic rate (A), effective quantum yield ((Dpsu), stomatal conductance (gs), and the ratio of intercellular to ambient CO2 (Ci/Ca), gs was corrected for temperature- induced changes in water viscosity and log-transformed prior to analysis. F- tests for fixed effects constructed by SAS MIXED procedure, with denominator degrees of freedom obtained from the Satterthwaite approximation and indicated in parentheses below each F-value. All random effects (population nested within elevation of origin, family nested within population, and their interactions with temperature) were estimated to be zero or near-zero and were not significant. Abbreviations as follows: Temp. = temperature, Spp. = species, Elev. = elevation. F for fixed sources of variation Trait Temp. Spp. Spp.*Temp. Elev.(Spp.) Elev.*Temp.(Spp.) df 1 1 l 2 2 A 2.13 0.12 27.03" 0.08 0.78 (3.71) (3.78) (3.71) (3.78) (3.69) cm 1.64 5.90’r 44.20*** 0.37 1.11 (56.5) (3.83) (56.5) (3.81) (55.7) gs 94.30"" 0.04 56.40"“ 1.35 4.30* (57.6) (5.24) (57.6) (5.21) (56.80) CilCa 66.59"“ 1.32 1629*” 2.99 6.91" (56.2) (4.80) (56.2) (4.78) (55.3) TP<0.10;"‘P<0.05; **P<0.01;***P<0.001;****P<0.0001 Table 12. P-values from single degree of freedom independent contrasts of least square means testing the null hypotheses that interspecific differences in physiological parameters within a temperature regime and intraspecific differences between temperature regimes are equal to zero. gs was corrected for temperature- induced changes in water viscosity and log-transformed prior to analysis. Intraspecific contrasts Interspecific contrasts cardinalis hot Iewisii hot cardinalis hot cardinalis cold Trait vs. cardinalis cold vs. Iewisii cold vs. Iewisii hot vs. Iewisii cold A 0.0123 0.0601 0.0201 0.0376 ¢psn 0.0002 <0.0001 0.0008 0.1309 gs <0.0001 0.1393 0.0004 0.0005 Ci/Ca <0.0001 0.0065 0.1560 0.0080 89 traits between range margin and range center populations is low. The random effects of population, family, and their interactions with temperature did not affect leaf physiological traits (data not shown). The species main effect in the model of effective quantum yield indicated that M cardinalis had a marginally higher light-adapted photochemical efficiency than M Iewisii. Both species attained higher net photosynthetic rates and effective quantum yields when grown under the temperature regime of their elevation range center, although the difference was only marginally significant for M Iewisii net photosynthesis (Figure 12a, b, Table 12). The main effect of temperature indicated that stomatal conductance and Ci/Ca were higher in the hot temperature regime than in the cold temperature regime. Greater conductance and Ci/Ca were detected at high temperature despite greater vapor pressure deficit (V PD) and lower relative humidity (RH) (M cardinalis: VPDhm, 2.13i0.05, VPDcold, 1.98d:0.05; RHho., 19.95i0.32%, RHcold, 31.39zt0.46%; M Iewisii: VPDhot, 3.06zt0.15, VPDcold, l.80:b0.05; RHhot, 14.35:]:0.45%, RHcold, 33.8721:0.59%). Mimulus cardinalis conductance was much lower in cold temperatures than in hot, whereas M Iewisii conductance was not significantly different between temperature regimes (Figure 12c, Table 12). Both species displayed higher Ci/Ca in hot than in cold temperatures, but M cardinalis showed a much larger decrease from hot to cold than M Iewisii (Figure 12d, Table 12). Within the hot temperature regime, M cardinalis displayed greater photosynthetic rate, effective quantum yield, and stomatal conductance than M Iewisii (Figure 12, Table 12). Within the cold temperature regime, M Iewisii displayed greater photosynthetic rate, stomatal conductance, and Ci/Ca than M cardinalis (Figure 12, Table 12). 90 10 l 0 8 E ,_‘ B — cardinalis “fir-x .. _2 "E ’- 3 . .. a 'm 8 >35 ()3 lewzsu 163’ E q 5 E 6 — LL. 0.6 — c: N 5 I 5‘ O a. 'E .9 2 4 0 E: 0.4 2: ° .2 H a. E 8 ”_ 8 3 2 2:11:12 .202 2 m 9 0 0.0 0.8 "N 1.0 o C 8 D 0 hi §TA06 — a 0'8 i 3.4” ' .2 'g 'E E 06 — 8 ONOA ~ __ g E g; 0.4 — <6 o '3 E 502 ’ E 0.2 — m 3 " 0.0 .5 0.0 ‘ ‘ Hot Cold Hot Cold Growth temperature Growth temperature Figure 12. Comparisons of species mean + SE physiological responses to low elevation (hot) and high elevation (cold) temperature regimes: a) net photosynthetic rate (A, in umol CO2 m'2 s"), b) effective quantum yield [(Dpsn =(Fm- — F s)/F m], c) stomatal conductance (gs, in mol H2O m' s", corrected for increased water viscosity at high temperature), and d) the ratio of intercellular to ambient CO2 (Ci/Ca). Populations of M cardinalis originating from the low elevation range center differed from populations originating from the mid elevation range boundary in the response of stomatal conductance and Ci/Ca to temperature. Low elevation populations showed greater decreases in conductance and Ci/Ca from hot to cold temperatures than mid elevation populations (data not shown), suggesting that mid elevation populations were more adversely affected by hot temperatures than low elevation populations or 91 were not as light-limited as low elevation populations in hot temperatures. However, no other M cardinalis traits and no M Iewisii traits displayed a pattern consistent with adaptive differentiation between range center and range margin populations. Within the cold temperature regime, M cardinalis maintained a significantly higher leaf temperature than M Iewisii (F 1,17.5=4.67, P=0.04). Although statistically significant, interspecific differences in leaf temperature within the cold temperature regime averaged only 0.6 °C and leaf temperature of both species was near ambient temperature. Within the hot temperature regime, M cardinalis maintained a significantly lower leaf temperature than M Iewisii (F1,43=38.02, P<0.0001). At high temperatures, high conductance enabled M cardinalis to maintain a leaf temperature approximately 10 °C below ambient, whereas M Iewisii leaf temperature was approximately 7 °C below ambient. Post-Freeze Tissue Damage Neither species was visibly damaged following the 0° C freeze. Individuals of both species were visibly injured by the -2° C freeze, but M cardinalis exhibited an average of 68.1% visible leaf tissue damage, whereas M Iewisii exhibited an average of only 46.3% damage (mixed model ANOVA: species, F1,55=14.77, P=0.0003). Whole Plant Performance Table 13 gives the results of linear mixed model analyses of stem length and log-transformed aboveground biomass. The main effect of temperature affected stem length but not aboveground biomass. The main effect of species and the interaction between species and growth temperature affected both traits. Elevation of origin, 92 Table 13. Linear mixed model analysis of variance summary for stem height and log- transformed aboveground biomass. F-tests for fixed effects constructed by SAS MIXED procedure, with denominator degrees of freedom obtained from the Satterthwaite approximation and indicated in parentheses below each F-value. All random effects (population nested within elevation of origin, family nested within population, and their interactions with temperature) were estimated to be zero or near- zero and were not significant. Symbols and abbreviations as in Table 11. F for fixed sources of variation Trait Temp. Spp. Spp.*Temp. Elev.(Spp.) Elev.*Temp.(Spp.) df 1 1 1 2 2 Length 489.41**** 64.78"" 219.02**** 8.32 1.03 (84.4) (27.4) (84.4) (27 .4) (84.4) Biomass 2.36 92.20MM 38.31 *** 0.07 1.85 (8) (8) (8) (8) (3) population, family and their interactions with temperature did not affect either growth trait. The main effect of temperature in the model of stem length indicated that plants were taller in the hot temperature regime than in the cold temperature regime. The main effect of species indicated that M cardinalis had greater stem length and aboveground biomass than M Iewisii. However, the interaction between species and temperature indicated that the species differed in grth response to temperature. Both species achieved greater stem length in hot than in cold temperatures, but the magnitude of size difference was much greater for M cardinalis than for M Iewisii (Figure 13a). Further, within the cold treatment, M Iewisii stem length was greater than M cardinalis stem length (Table 14). Although M cardinalis aboveground biomass was greater than M Iewisii biomass in both temperatures, M cardinalis aboveground biomass was greater in hot than in cold temperatures, whereas M Iewisii biomass was greater in cold than in hot temperatures (Figure 13b, Table 14). 93 3° 5 1 B — cardinalis A g CZE- Iewisii E o E V .9. "Eh .o ,3 'c 2 E a. 8 a o ..D < Hot Cold Hot Cold Growth temperature Growth temperature Figure 13. Species’ mean + SE a) stem length (cm) and b) aboveground biomass (g). Table 14. P-values from single degree of freedom independent contrasts of stem length and aboveground biomass least square means testing the null hypotheses that interspecific differences in growth parameters within a temperature regime and intraspecific differences between temperature regime are equal to zero. Intraspecific contrasts Interspecific contrasts cardinalis hot Iewisii hot cardinalis hot cardinalis cold Trait vs. cardinalis cold vs. Iewisii cold vs. Iewisii hot vs. Iewisii cold Length <0.0001 <0.0001 <0.0001 <0.0001 Biomass 0.0006 0.01 10 <0.0001 0.0423 Survival of both species was high in the cold temperature treatment (M cardinalis: 96.9%; M Iewisii: 93.8%), and survival of M cardinalis was 100% in the hot treatment. However, M Iewisii survival in the hot treatment was only 21.9%. DISCUSSION Interspecific Variation in Performance versus Temperature Mimulus cardinalis and M Iewisii displayed clear differences in performance under contrasting temperature regimes. Each species attained its greatest aboveground biomass when grown under a temperature regime characteristic of its altitudinal range 94 center and displayed reduced mass when grown under a temperature regime beyond its present altitudinal range. Although both species exhibited greater stem lengths in hot than in cold temperatures, the stem length of M cardinalis was more greatly reduced under cold temperatures than was that of M Iewisii. Survival of M Iewisii was also sensitive to temperature, showing a striking difference of 94% survival in cold temperatures and only 22% survival in hot temperatures. The low survival of M Iewisii in hot temperatures did not occur immediately upon exposure to high temperatures, but arose gradually throughout the experiment despite being well-watered. Plants appeared to waste away, implicating high respiration rates as the cause of reduced growth and survival (Hiesey et al. 1971). In hot temperatures, M cardinalis displayed greater survival, aboveground biomass and stem length than M Iewisii, whereas in cold temperatures, M Iewisii displayed greater stem length and resistance to freezing damage than M cardinalis. Previous studies have also demonstrated that M cardinalis and M Iewisii differ in growth response to temperature, although none have compared the species under natural temperature regimes (Cline and Agatep 1970; Hiesey et al. 1971). Hiesey et al. (1971) compared grth of M cardinalis from the foothills of the Sierra Nevada Mountains in California and M Iewisii from subalpine habitat in the Rocky Mountains of Montana under constant warm (30° C) or cold (10° C) temperatures and found that M Iewisii grew poorly under hot temperatures whereas M cardinalis was broadly tolerant of both hot and cold temperatures. Cline and Agatep (1970) grew Sierra Nevadan populations of each species (foothills M cardinalis, subalpine M Iewisii) under constant day and night temperatures of 3, 7, 11, 15, 19, 23, or 27° C. Both species 95 attained maximum growth at 19° C. However, M Iewisii experienced high mortality under hot temperatures but grew twice as fast as M cardinalis under cold temperatures. The magnitude of the difference in growth and survival between M cardinalis and M Iewisii was greater within the hot temperature regime than in the cold. Several factors may have played a role in producing the observed asymmetrical affect of temperature. First, in its natural habitat, particularly at mid elevations, M cardinalis is likely to experience occasional freezes and cool daytime temperatures late in the growing season, whereas M Iewisii is unlikely to encounter the extreme high temperatures used here anywhere within its natural range. Second, high light levels in a natural high elevation environment may exacerbate the effects of cold temperature by inducing photoinhibition (Close and Beadle 2003; Sayed 2003), but in our experiment light levels were relatively low, potentially moderating the harmful effect of low temperatures. Finally, in reciprocal transplant gardens at high elevation, mortality of M cardinalis is concentrated over the winter (Angert and Schemske, unpub. data). Because this experiment simulated conditions only during the growing season, it did not simulate the time period when M cardinalis is most susceptible to mortality. The patterns of differential growth and survival presented here are similar to those observed in reciprocal transplant gardens with these species at 415 m and 2395 m (Chapter 2), implying that temperature may be largely responsible for differences in grth and survival versus elevation. For example, after one growing season at 415 m, M cardinalis survival was 77% whereas M Iewisii survival was only 2%. In contrast, at the high elevation site, 2395 m, survival of both species was greater than 95% after one 96 growing season. Also, at high elevation, M cardinalis growth was reduced; seedlings were roughly two-thirds the size of M Iewisii after one growing season. Interspecific differences in growth response to temperature have been reported for several other congeneric species pairs differing in elevation distribution (Woodward and Pigott 1975; Woodward 1979; Graves and Taylor 1986; Woodward 1990; Kao et al. 1998). For example, growth of the low elevation species Sedum telephium, Dactylis glomerata, and Phleum bertolonii increases with temperature but growth of high elevation S. rosea, P. alpinum, and Sesleria albicans is insensitive to temperature (Woodward 1975; 1979). The differential sensitivity of S. telephium and S. rosea growth to temperature results in a switch in competitive dominance between low and high elevations (Woodward and Pigott 1975). Similarly, Graves and Taylor (1986) found that growth of Geum urbanum in cool temperatures was more restricted than grth of G. rivale, which occurs at higher elevations. However, in field experiments, the species exhibited only slight differences in relative growth rates across elevation. The results of these studies differ from ours in that the effect of temperature was more pronounced in low elevation species, supporting the generalization that lower range limits of high elevation species result primarily from biotic interactions such as competition rather than physiological limitation (MacArthur 1972; Woodward 1975; Scheidel et al. 2003). Instead, this study suggests severe abiotic limitation for M Iewisii beyond its lower elevation range limit due to inability to survive and grow under hot temperatures. Intraspecific Variation in Performance versus Temperature 97 To demonstrate that temperature limits species distributions requires the use of populations collected from range margins because marginal populations are often phenotypically or genetically divergent from more centrally located populations (Lesica and Allendorf 1995; Perez-Tris et al. 2000; Medail et al. 2002; Van Rossum et al. 2003; Faugeron et al. 2004) and may be differently adapted to temperature conditions at or beyond the range margin. However, in this study, populations from the range center and range margin of each species did not differ in grth or leaf physiological response to temperature, with the exception of M cardinalis for g5 and Ci/Ca. In reciprocal transplants at 415 m and 2395 m, a similar lack of adaptive differentiation with respect to elevation of origin was observed among populations of M cardinalis and M Iewisii (Chapter 2). Although finding no population differentiation in a controlled environment such as in the present study is consistent with results from the field, further experiments that simulate temperatures at the range margin are needed to investigate population variation in performance. The likelihood of population differentiation depends on the amount of gene flow as well as the degree of environmental difference between populations. Graves and Taylor (1988) also found no difference in the temperature acclimation of photosynthesis between populations of G. urbanum and G. rivale separated by only several hundred meters. Conversely, Pitterman and Sage (2000) found that a cold-acclimated low elevation population of Bouteloua gracilis exhibited depressed rates of net photosynthesis at cold temperatures but that a population originating 1500 m higher exhibited enhanced rates of photosynthesis at cold temperatures. Patterns of ecotypic differentiation in temperature response have also been found for T rifolium repens 98 photosynthesis in populations from 600 and 2040 m (Machler and Ndsberger 1977), for Eucalyptus pauciflora photosynthesis in populations from 915 and 1770 m (Slatyer 1977), and for Reynoutriajaponica growth in populations from 700 and 2420 m (Mariko et a1. 1993). Greater altitudinal separation between populations implies not only greater environmental difference but also greater geographic isolation. Populations used in this experiment originated at the elevation range center and range margin of each species, a difference in elevation of 600 -— 1200 m per species. Estimates of gene flow between range margin and range center populations of M cardinalis and M Iewisii would help determine whether gene flow prevents the evolution of local adaptation to the temperature conditions at range margins (Kirkpatrick and Barton 1997). lnterspecific Variation in Leaf Physiology versus Temperature Mimulus cardinalis and M Iewisii exhibit differences in leaf physiological response to temperature that are consistent with differences in growth response to temperature and with elevation distributions in nature. Each species attains the greatest net photosynthetic rate and effective quantum yield of PS 11 when grown under a temperature regime characteristic of its altitudinal range center and displays reduced photosynthetic rate and quantum yield when grown under a temperature regime beyond its present altitudinal range. Stomatal conductance and Ci/Ca are reduced under cold temperatures compared to hot temperatures, but M cardinalis shows much greater reductions than does M Iewisii. Within each temperature regime, the species native to that temperature exhibits greatest leaf physiological capacity. Differential sensitivity of M cardinalis and M Iewisii net photosynthetic rates to growth temperature demonstrates that each species is limited in its ability to acquire 99 primary resources when grown under a temperature regime beyond its elevation range. Hiesey et al. (1971) also demonstrated that M cardinalis and M Iewisii differ in photosynthetic response to temperature. When both species were grown at a constant temperature of 20° C, M Iewisii exhibited a light-saturated photosynthetic optimum that peaked at 25° C, but M cardinalis photosynthesis did not decline until temperatures exceeded 30° C. Contrary to our results, Graves and Taylor (1988) found little difference in the temperature response of photosynthesis between two species of Geum with different elevation distributions. The authors suggested that growth differences between the species were driven by differences in the ability to utilize assimilated carbon for growth, rather than by differences in the ability to assimilate carbon. Because photosynthesis is the primary source of energy and substrates for all other biosyntheses, when differences in photosynthetic rates are observed it is tempting to conclude that differences in carbon assimilation are directly related to differences in growth. However, although the observed differences in M cardinalis and M Iewisii carbon assimilation rates are consistent with their grth responses to temperature, instantaneous net photosynthetic rate is often a poor indicator of growth (Nelson 1988; Amtz et al. 1998). To fully dissect differences in growth requires measurement of respiration rates, plant architecture, and patterns of allocation in addition to measurement of photosynthetic rate. Future studies of M cardinalis and M Iewisii should identify how respiration, architecture and allocation vary between species and with temperature to further clarify growth limitations beyond the species’ elevation ranges. 100 Interspecific differences in the response of light-adapted quantum yield to temperature indicate that each species is able to use a larger fraction of incoming light energy for photochemical reactions when grown under the temperature regime of its elevation range center. Effective quantum yield is determined by the efficiency of excitation energy capture by open reactions centers and by the number of open reaction centers available for photochemical reactions (Schreiber et al. 1994). Decreases in the effective quantum yield of PS 11 may result from temperature-induced damage to electron transport processes or from feedback inhibition of PS 11 activity resulting from temperature-induced reductions in carbon metabolism (F alk et al. 1996; Laisk et al. 1998). To distinguish between these alternatives requires additional data on the temperature sensitivity of particular fluorescence parameters (e. g., minimum fluorescence, variable fluorescence, and non-photochemical quenching) in addition to detailed study of gas exchange metabolism (Owens 1994; Laisk et al. 1998; Xiong et al. 1999; Haldimann and Feller 2004). Without such information, it is difficult to attribute changes in fluorescence yield to any particular process (Owens 1994). However, studies of depression of net photosynthesis in oaks (Haldimann and Feller 2004) and Antarctic plants (Xiong et al. 1999) have concluded that heat-induced damage to thylakoid membranes does not occur until temperatures well above those that depress photosynthesis, and thus that reduced enzymatic activity is the main cause of depressions in photosynthesis under high temperatures in the field. Likewise, low temperature may harm photosynthesis primarily through effects on carbon metabolism rather than effects on photochemistry (Leegood and Edwards 1996). 101 Patterns of variation in stomatal conductance and CglCa differed from patterns for photosynthetic rate and effective quantum yield. In hot temperatures with high vapor pressure deficit and no water limitation, M. cardinalis showed high stomatal conductance, which allowed greater evaporative cooling of the leaf surface. Even with no water limitation, M Iewisii showed lower stomatal conductance under hot temperatures than M cardinalis, higher leaf temperatures, and lower intercellular concentrations of CO2. High Ci/C8| ratios in hot temperatures, particularly of M cardinalis, indicate that photosynthesis at high temperatures was possibly light limited. However, subsequent experiments using higher light levels during growth and measurement find similar patterns of difference in photosynthetic rates between species and between temperature regimes (A. Angert, unpub. data), and it is unlikely that greater light levels would have eliminated the observed differences between M Iewisii and M cardinalis in the hot temperature regime. Without measurement of the CO2 saturation point for photosynthesis, it is unclear whether lower stomatal conductance for both species, particularly M cardinalis, in cold temperatures resulted in greater stomatal limitation to photosynthesis. However, it is likely that lower conductance resulted from, rather than caused, low photosynthesis. Long—term acclimation to growth temperature and light conditions during our study may have allowed changes in stomatal density or aperture that optimized conductance to reduce unnecessary transpiration in conditions of low CO2 assimilation (Ferris et al. 1996). Several other studies support this hypothesis. Naidu and Long (2004) found that cold-acclimated Zea mays did not experience increased stomatal limitation to photosynthesis, despite greatly reduced stomatal 102 conductance. Similar results have been reported for tomato (Martin and Ort 1985), olive (Bongi and Long 1987), rye (Huner et al. 1986), wheat (Hurry and Huner 1991), and several C4 grasses (Pitterman and Sage 2001; Naidu and Long 2004). As in these examples, it is likely that conductance decreased to match assimilatory use of CO2 and that reduced intercellular concentrations of CO2 resulted from, rather than caused, low photosynthetic rates. Some of the observed physiological responses may be due to uncontrolled environmental variables that covaried with temperature, such as vapor pressure deficit or relative humidity, rather than temperature per se (Matzner and Comstock 2001). However, increased conductance was observed at high temperatures despite greater vapor pressure deficit and reduced humidity. Further, although these factors may be confounded in the present study, this represents a realistic natural scenario in temperate environments, where temperature and vapor pressure deficit often increase simultaneously (Iio et al. 2004). This study provides evidence that M cardinalis and M Iewisii differ in performance under temperature regimes characterizing their contrasting low and high elevation range centers. Differences in the species’ leaf physiological responses under contrasting temperature regimes are consistent with differences in performance observed in both controlled and natural environments. Elevation range limits of M cardinalis and M Iewisii may arise, in part, due to metabolic limitations on growth that ultimately decrease survival and limit reproduction. 103 CHAPTER 4 Natural selection within and beyond the elevation ranges of monkeyflowers (Mimulus cardinalis and M. Iewisii) Abstract.— Every species occupies a restricted geographic distribution, but why natural selection does not increase tolerance to limiting environmental variables and allow continual range expansion remains an evolutionary conundrum. At the heart of many of hypotheses for distribution limits is the idea that environments within and beyond the species range select for different phenotypes, and it is the difficulty of producing a phenotype adapted to environments both within and beyond the range that constrains range expansion. In this study, I examine natural selection in sister species of monkeyflower, M cardinalis and M Iewisii, to identify traits that contribute to fitness within and beyond elevation range limits and to ask whether adaptation to environments beyond the range entails a cost to adaptation within the range. I transplanted interspecific hybrids to low and high elevation and found that selection favored early flowering at high elevation. Hybrids selected at low elevation displayed increased leaf photochemical efficiency in warm temperatures. Selection acted in the direction of the native parental species’ trait value, supporting the hypothesis that M cardinalis photosynthetic traits and M Iewisii flowering phenology are adaptive at their respective low and high elevation range centers. If adaptation to one environment entails a cost to adaptation in other environments, then selected hybrid populations should display reduced fitness, relative to an unselected control population, when grown in an environment in which they were not selected. One such tradeoff was observed in this 104 study, where hybrids selected at high elevation displayed reduced biomass when grown in temperatures characteristic of low elevation. Continued generations of experimental evolution and the reciprocal transplantation of selected populations to low and high elevation will allow definitive tests of the role of between-environment fitness tradeoffs in range limit evolution. Key words: natural selection, experimental evolution, range limits, phenology, physiological adaptation Species’ distribution boundaries have long fascinated ecologists and biogeographers seeking explanations for why species fail to occur beyond their present limits (Griggs 1914; Grinnell 1917; Good 1931; Dahl 1951). Most studies of distribution limits have focused on identifying the proximate ecological factors that give rise to a distribution boundary. Such studies may examine populations, asking whether local abundance decreases towards the range margin (Brown et al. 1996; Sagarin and Gaines 2002) or whether marginal populations are demographic sinks or more prone to extinction than central populations (Carter and Prince 1981; Lennon et al. 1997; Mehlman 1997; Guo et a1. 2005). Many other investigations of distribution limits focus on individuals, asking whether survival and reproduction decrease towards the range margin (Marshall 1968; Pigott and Huntley 1981; McKee and Richards 1996; Garcia et al. 2000; Hennenberg and Bruelheide 2003), and, if so, which environmental variables are responsible for variation in components of fitness (McNab 1973; Root 1988a; Cumming 2002). However, even when satisfactory answers to these questions are found, a central question remains: why does natural selection not continually improve 105 adaptation to limiting environmental variables and overcome current distribution limits? To answer this question, we must know which traits are under selection at and beyond the range boundary, and why they do not evolve to allow range expansion. Many mechanisms have been proposed to limit the potential for adaptive evolution at and beyond range boundaries, including a lack of genetic variation in fitness-related traits (Antonovics 1976; Bradshaw and McNeilly 1991), an influx of maladapted genotypes from populations at the range center Hialdane 1956; Kirkpatrick and Barton 1997), and negative genetic correlations either among fitness-related traits or between fitness in environments within versus beyond the range (Antonovics 1976; Holt 2003). At the heart of many of these hypotheses is the idea that environments within and beyond the range select for different phenotypes, and it is the difficulty of producing a phenotype adapted to environments both within and beyond the range that constrains range expansion. In this study, I examined natural selection in sister species of monkeyflower, M cardinalis and M Iewisii, to identify traits that contribute to fitness within and beyond the elevation range limit and to ask whether adaptation to environments beyond the range entails a cost to adaptation within the range. Previous experiments have demonstrated that each species is most fit at its elevation range center (low elevation for M cardinalis, high elevation for M Iewisii), less fit at the shared mid-elevation range boundary, and unable to survive and reproduce when transplanted to elevations beyond its current range (Chapter 2). For M Iewisii, reduced fitness at low elevation results primarily from high mortality within the first growing season. For M cardinalis, 106 reduced fitness at high elevation is due primarily to limited growth and reproduction (Chapter 2). Many features of the environment that affect plant survival, growth, and reproduction change with elevation, most prominently temperature and length of growing season. In growth chamber experiments, Mimulus cardinalis and M Iewisii display differences in survival, growth, leaf physiology, and freezing resistance under temperature regimes that mimic their contrasting low and high elevation range centers (Chapter 3). The species also differ in phenological traits that may contribute to differences in fitness versus elevation. When grown in a common environment, M Iewisii flowers earlier than M cardinalis (Hiesey et al. 1971), suggesting that the ability to flower quickly and mature fruits rapidly may be favored in short growing seasons at high elevation. In this study, I measure natural selection on leaf physiology, freezing resistance, and flowering phenology. I hypothesize that these traits affect the ability to survive and reproduce at different elevations. One major difficulty in measuring natural selection across species’ ranges is that populations only exist above some threshold of fitness, limiting the environmental axis along which we can measure selection in natural populations. Also, trait variation within populations is likely to be minimized by stabilizing selection (Endler 1986) and influenced by factors other than natural selection, such as phylogenetic history (Harvey and Pagel 1991). These difficulties call for experimental approaches to detect selection. To increase the range of trait variation available to natural selection and to create trait combinations not found within either species, I created late-generation hybrids between M cardinalis and M Iewisii and transplanted them to low and high elevation. After one 107 generation of evolution in each environment, I compared phenotypes between selected hybrid populations and an unselected greenhouse control population to identify traits that showed a shift in mean value after selection. I apply two criteria to assess trait evolution. First, if a particular trait is itself a target of natural selection or is genetically correlated with a trait that is the target of natural selection, then its mean value should differ significantly from the unselected control population. Second, if parental trait values are adaptive, then selected hybrid trait means should evolve in the direction of the parent native to that environment. Based on these criteria, I hypothesize that hybrids selected at high elevation will flower more rapidly, exhibit less tissue damage following freezes, and display greater leaf physiological capacity in cool temperatures characteristic of high elevation than the greenhouse control population. Likewise, I hypothesize that hybrids selected at low elevation will display greater leaf physiological capacity in warm temperatures characteristic of low elevation than the greenhouse control population. To determine whether adaptation to low elevation entails a cost to adaptation at high elevation, and vice versa, I measured phenotypes on hybrids grown in two temperature regimes: one characteristic of low elevation and one characteristic of high elevation. If adaptation to one environment entails a cost to adaptation in another environment, then selected hybrid populations should display reduced fitness, relative to the control, when grown in the environment in which they were not selected. Alternatively, if reduced fitness in the unselected environment is not evident as a pleiotropic byproduct of evolution in the selected environment, then we can conclude 108 that fitness within each environment is able to evolve independently and between- environment fitness trade-offs are not present. MATERIALS AND METHODS Study system Mimulus cardinalis and M Iewisii (Phrymaceae) are perennial herbs of riparian habitats in western North America. Mimulus cardinalis occurs from southern Oregon to northern Baja California, Mexico and from the coast of California inland to Arizona and Nevada. Mimulus Iewisii is composed of two partially incompatible races, one occurring in the Pacific Northwest and the Rocky Mountains and one occurring primarily in the Sierra Nevada Mountains of California (Hiesey et al. 1971; Hickman 1993; Beardsley et al. 2003). In California, M cardinalis and M Iewisii occupy different elevation ranges, with M cardinalis occurring from sea level to 2400 m and M Iewisii occurring from 1200 m to 3100 m (Hickman 1993). In the Yosemite National Park region where this research was conducted, M cardinalis is not found above 1500 m, M Iewisii is not commonly found above 2800 m, and the species may co-occur on larger watercourses between 1200 and 1500 m elevation. Generation of hybrid populations for transplants Seeds of M cardinalis and M Iewisii were collected from a naturally occurring sympatric population along the South Fork of the Tuolumne River (Carlon Day Use Area, Tuolumne County, California, 1320 m) in September 1999. Two individuals of each species from distinct maternal plants were grown to flowering in the University of Washington greenhouse under standard greenhouse conditions and crossed to generate two independent Fl hybrid lines, using the same species (M Iewisii) as the maternal 109 parent in each cross. Two F1 individuals, one from each line, were grown to flowering and crossed to generate an F2 population. One thousand F2 individuals were grown to flowering and crossed to one another so that each plant served as pollen donor and recipient once (with no self- or reciprocal pollinations), generating 1000 hybrid seed lots with an additional round of recombination. Transplant gardens Experimental gardens were established near Jamestown, California (415 m) and at White Wolf Ranger Station in Yosemite National Park (2395 m; Figure 6). These locations were chosen to represent elevations for each species that are central within the elevation range (415 m for M cardinalis, 2395 m for M Iewisii) and beyond the range boundary (2395 m for M cardinalis, 415 m for M Iewisii). Seeds from 500 hybrid seed lots were sown in flats in the University of Washington greenhouse five weeks prior to transport to garden sites. The average age of transplanted seedlings was approximately three weeks after germination. In July 2001, 8110 seedlings (16-17 individuals from each of 500 seed lots) were transplanted in random order at 2395 m. To assess the strength of selection in each environment, 319 seedlings of each parental species were randomly interspersed among the hybrid individuals. In April 2003, seedlings were transplanted to 415 m following identical methods, except that space limitations allowed only 6000 individuals (1 1-12 from each of the same 500 hybrid seed lots plus 156 of each parent) to be transplanted. Garden plots were covered in landscape fabric and irrigated daily to approximate conditions in the species’ native riparian habitat and to standardize water treatments across environments. Due to irrigation system failure in one area of the Jamestown garden, 27 M cardinalis, 24 M Iewisii, and 933 hybrids 110 were excluded from analysis. Survival and day of first flowering were recorded from 2001-2003 at White Wolf (2395 m) and in 2003 at Jamestown (415 m). After only one growing season at Jamestown, most M Iewisii were dead and the majority of surviving plants had reached the flowering stage. Observations were conducted over a longer time period at White Wolf because of the length of time necessary for plants to reach reproductive maturity at high elevation. Data were recorded at approximately two-week intervals throughout each growing season. Generation of selected and control hybrid populations Because some phenotypes of interest were impractical to measure in the field, I generated a selected seed population at each elevation for trait measurement under controlled conditions. Selected seed populations were made by crossing subsets of individuals that were able to survive and flower within the transplant gardens. At White Wolf, pollinations were conducted at two-week intervals in 2003, beginning two weeks after flowering commenced and proceeding throughout the flowering period. Up to 80 individuals were crossed to one another within each pollination cohort, using only those individuals that began flowering within the interval. Buds were enclosed in mesh bags to prevent pollinator visitation. Each plant served as pollen donor and recipient only once. When more than 80 individuals began flowering within the two-week period, individuals were haphazardly selected from throughout the garden. Because this method of crossing potentially flattened the flowering time distribution of the offspring, for subsequent experiments I included fruits from each pollination cohort in proportion to the total number of individuals that began flowering within the interval. At Jamestown, pollinations during the growing season of 2003 were unsuccessful, so dormant rhizomes lll of individuals that survived and flowered in 2003 were transported to the Michigan State University greenhouse in February 2004, where plants were regrown to flowering. Pollinations of Jamestown plants grown in the greenhouse were conducted following identical methods to those used at White Wolf, defining pollination cohorts by the flowering times recorded within the transplant garden. An additional population of hybrids from 250 of the original 500 hybrid seed lots was grown under favorable conditions in the greenhouse, where selection was assumed to be minimal (survival 100%, only 6 out of the initial 250 lines not included in crosses due to pollen sterility), and crossed to generate a control population of hybrid seeds. Measurement of phenotypic traits To examine the relationship between flowering time and seed set at White Wolf, seed set per fruit was quantified for hand pollinations conducted at two-week intervals (see Generation of selected and control hybrid populations above). In the lab, samples of approximately 150-200 seeds per fruit were counted under a dissecting microscope and weighed to determine the relationship between seed mass and seed number. Seed number per fruit was then estimated from the total seed mass. All other phenotypic traits were measured in growth chambers on two sets of selected and control hybrids, one grown in a temperature regime characteristic of low elevation and a second grown in a temperature regime characteristic of high elevation. Low and high elevation temperature regimes were based on July temperatures recorded within the Jamestown and White Wolf transplant gardens (detailed in Chapter 3). The low elevation chamber was set for a 35 °C daytime maximum and 15 °C nighttime minimum, with 42 °C daytime maximums on days 50 and 64. The high elevation 112 chamber was set for a 23 °C daytime maximum and 4 °C nighttime minimum, with two -2 °C freezes on nights 50 and 64. Daily maximum and minimum temperatures were held for four hours each with gradual ramps between maximum and minimum temperatures. Two identical plant growth chambers were used for low and high elevation temperature regimes (Model GC-20BDAF-REFR404, Econair, Winnipeg, Canada) except during freezing events, when plants were transferred for a period of 24 hours to a chamber capable of holding sub-zero temperatures (Model GC-20BDAF- REFR-22, Econair, Winnipeg, Canada). Chambers were programmed for 14/10 hour day/night cycles. Light averaged 350 umol photons m'2 s'1 at plant height during the daytime period. In October 2004, seeds of selected and control hybrid populations were sown in either the low or the high elevation temperature regime. Pots were placed in random order within wire frames, and the frames were placed in trays for sub-irrigation within the growth chamber. Frames were rotated several times per week to minimize position effects. Approximately 10 seeds were sown per 10 cm pot and seedlings were randomly thinned to one seedling per pot three weeks after sowing so that each temperature regime contained 35 individuals from each hybrid population plus 15 individuals of each parent species. After thinning, the cotyledon diameter of each remaining seedling was measured to account for potential differences in performance between selected populations due to maternal growth environment (greenhouse or 2395 m garden). However, cotyledon diameter did not differ between selected populations (one-way analysis of variance, low elevation temperature regime: F2,163=0.12, P=0.89; high 113 elevation temperature regime: F2,155=1.01, P=0.37), indicating that seed quality did not measurably influence early seedling growth. Simultaneous gas exchange and chlorophyll fluorescence measurements were performed following the last extreme temperature event for each treatment to characterize leaf photosynthetic function in low and high elevation temperature environments. Measurements were conducted as described in Chapter 3. Four leaf physiological traits were quantified: l) instantaneous net photosynthetic rate (umol CO2 In2 S"); 2) effective quantum yield of photosystem II [(Fm’ — Fs)/ Pay], the fraction of absorbed photons that a light-adapted leaf uses for photochemical reactions, determined by chlorophyll fluorescence readings; 3) stomatal conductance (mol H2O m'2 s"), an indicator of the degree of stomatal openness, which determines leaf loss of water and gain of carbon dioxide; and 4) the ratio of intercellular to ambient CO2, which can indicate the degree to which stomatal closure limits the availability of CO2 for photosynthesis. These variables may indicate particular processes that limit photosynthetic activity in suboptimal temperatures and, together, summarize leaf photosynthetic function. To quantify post-freeze tissue damage, I estimated the percentage of total leaf tissue damaged on each plant on the day following each freeze event within the low elevation temperature regime. The date of first flower was recorded for every flowering plant. After 87 days (low elevation temperature regime) and 127 days (high elevation temperature regime), plants were harvested for measurement of aboveground biomass, length of the main stem, and flower number. The difference in time period preceding harvest reflects large differences in growth rates between temperatures. Despite additional time in the growth chamber, very few 114 plants (1 M Iewisii, 7 hybrids) reached the flowering stage within the high elevation temperature regime. For this reason, flowering phenology and flower number were compared within the low elevation temperature regime only. Data analysis Performance and reproductive phenology in transplant gardens.— To examine differences among parents and hybrids in the probability of surviving and flowering at each elevation, I performed logistic regressions (PROC LOGISTIC, SAS Institute, Cary, NC, USA), using the ‘contrast’ statement to test for pairwise differences between parents and hybrids and the sequential Bonferroni procedure to control type I error rates. To examine differences among parents and hybrids in the day of first flower at each elevation, I performed analysis of variance (ANOVA) on log-transformed data (PROC GLM, SAS Institute, Cary, NC, USA). Pairwise differences between parents and hybrids were evaluated with Tukey—Kramer adjusted comparisons of least square means. To examine the relationship between flowering time and seed set at White Wolf, linear regressions of seed count per fruit versus pollination date were performed (PROC REG, SAS Institute, Cary, NC, USA). Separate linear regression analyses were conducted for M Iewisii and hybrids; small sample size of flowering M cardinalis (N=2) prevented analysis of seed set data for this species. Performance, physiology, and phenology in growth chambers.— Two sets of analyses were performed. I first analyzed data from parental species to verify interspecific differences and assess the effect of each temperature regime, and then analyzed hybrid populations to test for evolved differences after natural selection. Variables were analyzed as one-way designs to assess the effect of species (or hybrid 115 selection regime) within each temperature environment. One-way designs were chosen because different variables were analyzed in each temperature environment (e. g., freezing damage only in the high elevation temperature regime, reproductive variables only in the low elevation temperature regime), precluding two-way designs that included the effect of temperature. To examine differences in survival among hybrid populations, I used logistic regressions as described above; no mortality of M cardinalis prevented analysis of interspecific differences in survival. For all other variables, univariate ANOVA tests were performed using PROC GLM (SAS, SAS Institute, Cary, NC, USA). Multivariate approaches were not chosen because the overall goal was to assess the effect of selection regime on each dependent variable singly, rather than to understand the effect of selection regime on the multivariate distribution of response variables (Huberty and Morris 1989). Care was taken to select independent or only weakly correlated variables and to control the type I error rate with sequential Bonferroni adjustments. To examine pairwise differences between hybrid populations, I used Tukey-Kramer adjusted comparisons of least square means. RESULTS Performance in reciprocal transplant gardens At low elevation, survival of M cardinalis was high (81%) and nearly every surviving plant flowered in the first growing season (Table 15). In contrast, survival of M Iewisii at low elevation was very low (17%), and fewer than half of all surviving plants flowered. At high elevation after three growing seasons, M cardinalis survival was low (7%) whereas M Iewisii survival was much higher (41%). Only two M cardinalis plants flowered at high elevation, whereas approximately two-thirds of 116 58.? 002.00 88.? 28.2 a seguemaaé 88.? 89.2 88.? 804.2 _ 0.2.02.2 .9 205280 580v $3.2 88.9 024.2. a 652 .0, 022280 soodv 28.2: :0de 22.2 m 252.00 203 333 58.9 0202 soodv 3.0.; _ 322 03.2200 58.9 22.00 $8.9 $8.8 _ 0.222 a, 425280 88.0 $2.2 $8.0 28.2 a 832 .0, 025280 88.? ~58 soodv 33:: N 80800 caeaesé 0 xx 0 xx 2 280800 @0530; 33056 000mm 00080 .0553?“ €800.33“ 30000000 000% “000$:me £080.— 0000000b€ =< 00300—0 A8 mdmN £03 00:56 nwE 00.0 A8 m :4 .03020803 32 H0 030% 02.5.3 30000000.: 000 4.0233 3 £23380 .3 00m wctoaoc m0 bzfimnofi 000 .0333 .00 tax—53000 05 .«0 002000th oamMmS .2 050B 2 mm 82 e; 38 00am 2 s. 2 3 e: 322 2 as“; a _ e a m an 02528 .2 be? 323 me we 3 gm 2:. Ea: m S w 0 3 0.222 .2 can; 0 E S m 8. 02 4.2528 .2 geese. a0>m§0m0d 805305 0:538 .x. 33:3 .5 022 z 032 z Bag: 2 090800 8030 .3808 950% 0005 00.00 .203 0033 H0 98 000000 wEBBw 0:0 00% 030000803 00 00000000 800 £80300 £58582 2562 820 30:8 05 a 8% 0e 38 .033 2230 8250 :02 2a 0: m 2 £398,500 00000—0 32 00 02.53 0500000003 000 006000 3:030 m0 wat0>>0a 000 3330.01 .2 030,—. 117 surviving M. Iewisii flowered in the third growing season. Within each garden, survival and flowering of hybrids was intermediate to the parents. Logistic regressions of the probability of survival and flowering confirm that, within each garden, the species native to that elevation was more likely to survive and flower than either the non-native species or hybrids (Table 16). Relative fitness of parents and hybrids within each garden was calculated by dividing the proportion of plants surviving to flower by the proportion observed for the species native to that elevation (Figure 14). At low elevation, hybrid relative fitness was approximately 0.8, whereas at high elevation, hybrid relative fitness was approximately 0.4, indicating stronger selection against hybrids at high elevation than at low. _ M cardinalis 1.0 ‘ E22223 Hybrid _ 1:1 M Iewisii 3’; 0.8 - <1) 5 ‘1; 0.6+ .2 E 04 £2 ' “ 0.2 e 0.0 Low High Elevation Figure 14. Relative fitness of parental species and hybrids transplanted to low (Jamestown, 415 m) and high (White Wolf, 2395 m) elevation. 118 Reproductive phenology The day of first flower differed significantly among parents and hybrids at both low elevation (ANOVA, F2,3og3=6.54, P=0.0015) and high elevation (ANOVA, F290 1:41 .63, P<0.0001) in 2003. At low elevation, M. cardinalis flowered on average 4 days later than hybrids (t3033=3.60, Tukey-Kramer adjusted P=0.0009). At high elevation, M cardinalis and hybrids flowered significantly later than M Iewisii. On average, hybrids flowered approximately 13 days after M Iewisii (t9m==2.87, Tukey- Kramer adjusted P=0.0118), and the two M cardinalis to flower did so approximately 35 days after M Iewisii (t901=8.98, Tukey-Kramer adjusted P<0.0001). Although all plants at high elevation flowered approximately one week later in 2003 than in 2002, relative differences among hybrids and parents were similar in both years (data not shown). At high elevation, seed number per fruit declined linearly with pollination date 3000 3: 250° ‘ o . Hybrid. .. E O M lewzsu H 2000 - 0 8. O o 33 1500 .0 o g 1000 - M. Iewisii c: b = -22.08* '93 500 - "’ 0 - Hybrid .- . b = -16.75*.** 190 200 210 220 230 240 250 260 Julian day of pollination Figure 15. Seed number per fruit versus pollination date for M Iewisii and hybrids grown at high elevation (White Wolf, 2395 m). Regression coefficients (b) from linear regression analysis (* P < 0.05, *** P < 0.001). 119 Stem length (cm) Instantaneous photosynthetic rate (umol C02 m"2 s'l) Effective quantum yield [(Fv' ' 1:sva'] 60 50 _ A 1120:1719 P=0.0005 40 - 30 ~ 20 e 10 0 5': j, ]-:n M card. M lew. 14 12 _ C FU9=50.10 10 - P<0.0001 8 _ 6 _ 4 _ 2 a 0 _ .. , -..i M card. M lew. 0.8 E F1,19=15.20 0.6 4 P=0.0010 0.4 - 0.2 - 0.0 . " M card. M lew. Stomatal conductance (mol H20 In"2 s'l) Intercellularzambient C02 Aboveground biomass (2:) 0‘1 Fm=68.86 P<0.0001 m M card. M lew. 1.4 1.2 ~ 1.0 ~ 0.8 1 0.6 - 0.4 - 0.2 - 0.0 - Fl’19=35.53 P<0.0001 M card. M lew. 1.00 0.95 - 0.90 - 0.85 - FU,=7.49 P=0.0131 0.80 M card. M lew. Figure 16. Mean (+ SE) trait values of M cardinalis and M Iewisii in the low elevation temperature regime. a) stem length, b) aboveground biomass, C) instantaneous net photosynthetic rate, d) stomatal conductance, e) effective quantum yield, and f) intercellular: ambient C02. Results of one—way ANOVA testing the effect of species given for each trait. All values remain significant after sequential Bonferroni adjustment to maintain a type I error rate of 0.05. 120 25 A F1’26=1.47 a 20 " P=0.2367 C: A 15 ‘ 2 .8. EV 10 _ (I) 51 0 , I ‘ +3 M card. M lew. H .-.}—j A 18 4:: 'T 16 — F,19=0.14 ‘a’ "’ ' he? 14 v P=0.7150 f’g’ a 12 ~ 0 N 10 '5. 8 81 (I: :3 '3 6 ‘ GE 8 4" s 5': 2 ~ . g 0 ’r... 4,. g M card. M lew. “O E) 0.8 a E F1,19=0'44 N: P=0.5135 5;; 0.6 ~ ,, 5 2’ g, ' 04 . o ’> .5 9;, 0.2 - 8 _. . , E 0.0 - '- ‘ - ' M card. M lew. Stomatal conductance Aboveground biomass (g) (mol H20 rn'2 s'l) Intercellularzambient C02 14 - Fl,25=72'15 12 - P<0.0001 10 ~ 8 _4 6 _ 4 .4 2 _ |_‘|'j 0 I l M card. M lew. 0.4 F,_,,=0.00 0,3 . P=0.987l 0.2 — 0.1 - 0.0 — ' ' ' M card. M lew. 1'2 F 0 39 _ 1,19= ' 1'0 P=O.5399 0.8 - F... 0.6 - 0.4 - 0.2 - . y .4! 0.0 - ' “i“ M card. M lew. Figure 17. Mean (+ SE) trait values of M cardinalis and M Iewisii in the high elevation temperature regime. A) Stem length, B) aboveground biomass, C) instantaneous net photosynthetic rate, D) stomatal conductance, E) effective quantum yield, and F) intercellular: ambient C02. Results of one-way ANOVA testing the effect of species given for each trait. The difference in biomass remains significant after sequential Bonferroni adjustment to maintain a type I error rate of 0.05. 121 for both M Iewisii (b = -22.08, N = 24, t=-2.30, P=0.0311) and hybrids (b = -16.75, N = 149, t = 3.41 , P = 0.0008), indicating selection for early flowering at high elevation (Figure 15). Phenotypic differences within growth chambers Parental species.—In the low elevation temperature regime, M cardinalis survival was 100%, and 92% of individuals reached the flowering stage, whereas M Iewisii survival was only 36% and no individuals flowered. Within the low elevation temperature regime, M cardinalis exhibited greater growth and leaf physiological capacity than M Iewisii (Figure 16). Within the high elevation temperature regime, survival of both species was high (M cardinalis, 100%; M Iewisii, 92%) and few individuals flowered (M cardinalis, 0%; M Iewisii, 9%). Mimulus cardinalis and M Iewisii did not show significant differences in most measured traits (Figure 17), although M cardinalis again attained greater biomass. Although not statistically significant, M Iewisii had slightly greater rates of photosynthesis, effective quantum yield, and intercellular C02 concentration than M cardinalis (Figure 17) and numerically less tissue damage following freeze events than M cardinalis (Freeze 1: M cardinalis, 27.1 i 8.2%, M lewsii, 15.0 :1: 6.8 %, F1,25=0.82, P=0.3735; Freeze 2: M cardinalis 19.2 :1: 8.8%, M Iewisii, 9.2 :1: 7.4%, F1,23=1.03, P=0.3201). Hybrid p0pulati0ns.— High correlations were detected for two pairs of leaf physiological traits: instantaneous photosynthetic rate and effective quantum yield, and stomatal conductance and the ratio of intercellular to ambient C02 (Table 17). For this reason, photosynthetic rate and the ratio of intercellular to ambient C02 were excluded from analyses examining the effects of selection regime on trait evolution. Effective 122 «E «E S: Q: Q: «E Am 885 038% .x. :3 85 s: e: we «a e: A _ 385 ”mafia as $8 $8 38 as we Q: :25 :5 ...Rd :42 53o: 3 a3 38 Sc 3% Ga Ge 85 8.? 8: mod tame 2.236 2% 83530 $8 Sc 88 Ge 63 3 .o 3.3 s: :5 ...ttfid :.o Nou eases £8.35 9% Se 88 38 3% mod :6 «\c “willed :EFwd 1:.vmd oozfioaficoo EmEoum as Ea 88 as 38 3o- 85 a: 3:35 1....on timed ea savages: AN @305 C 38.5 832m 20% ~00 323:8 8:826:00 88 owwg .X. @9258 o\.. 9 PEG 835:0 2:862: EmEBm ovofigmowonm .Eomowtooo :03 33.3 mumofiaewm 3 53w 3% macaw .2806 v 9.21.... .Sod v 9.2.2.. .56 v 9.2.. .36 v m... .36 v mt $56508 noun—oboe cembwom 8.3%:me 8865 3333‘ 358:3“ Ho: u 23:: 3 36225 25on 0558388 gage—o 32 2t 5 3&0on 8: $3 omega 38¢ 98 0::on 238388 cargo—o :wE 05 E 33036 8: 33 warez/2m .Ecowmmw 05 322— con 25on 288%an 2235—0 32 05 5‘23 80:22.80 .38me 2: 959m 53m 25on 288388 gouge—o SE: 05 ESE» 28328.80 $3820 53on E 8::on 058388 .8396? :wE use 32 a :3on €an co 3.5808 33.: mo mourns:— couflohoo .2 2an 123 Table 18. Analysis of variance summary for physiological and phenological traits and fitness components of hybrids grown in a low elevation temperature regime. For each variable, the effect of selection regime (low elevation, high elevation, or greenhouse control) was tested with one-way ANOVA. Values in boldface remain significant after sequential Bonferroni correction (Rice 1989). Category Response df num. SS MS F P (den) (SSE) (MSE) Trait Conductance 2 0.6 0.3 4.44 0.0144 (92) (5.8) (0.1) Quantum yield 2 0.04 0.02 8.99 0.0003 (92) (0.2) (0.002) Days to flowering 2 418.9 209.5 2.14 0.1254 (70) (6854.1) (97.9) Fitness Stem length 2 608.4 304.2 1.28 0.2823 component (98) (23266.2) (237.4) Biomass 2 174.8 87.4 6.03 0.0034 (98) (1421.5) (14.5) Flower number 2 386.6 193.3 0.19 0.8237 (98) (97463.7) (994.5) Table 19. Analysis of variance summary for traits and fitness components of hybrids grown in a high elevation temperature regime. For each trait, the effect of selection regime (low elevation, high elevation, or greenhouse control) was tested with one-way ANOVA. Stem length was log-transformed and percent tissue damage was arcsine square-root transformed prior to analysis. No values remain significant after sequential Bonferroni correction (Rice 1989). Category Response df num. SS MS F P (den) (SSE) @ASE) Trait Conductance 2 0. 1 0.02 1 .68 0. 1922 (83) (1.1) (0.01) Quantum yield 2 0.01 0.002 0.19 0.8259 (83) (1.3) (0.02) % Damage (freeze 1) 2 1.6 0.8 3.93 0.0228 (95) (19.2) (0.2) % Damage (freeze 2) 2 0.16 0.1 0.57 0.5652 (89) (12.19) (0.1) Fitness Stem length 2 0.8 0.4 0.69 0.5054 component (94) (56.6) (0.6) Biomass 2 68.6 34.3 0.81 0.4467 (93) (3922.0) (42.2) 124 '2 08 _ A o 1.2 ‘ B F2_92=4.44 .am :5 .1." 1.0 _ P=0.0144 g ”E 0.6 — g N” b E u ' 0.8 ‘ 5’9 8 E a”: 0.4 — o % 0.6 « g E '5 :1: 'SE a '3‘ 0.4 ‘ 8 0.2 - o E E .2 V 0.2 - 0.0 - “ “ ‘ 'i‘ 0.0 . - . low control high low control high Hybrid selection regime Hybrid selection regime 80 C F2,70=2.l4 g P=0.1254 a 60 1 E I... 8 40 ‘ _.’_‘ “(a :‘ E 20‘ , i low control high Hybrid selection regime Figure 18. The effect of selection regime (low elevation, greenhouse control, or high elevation) on two leaf physiological traits and flowering phenology of hybrids grown in a temperature regime characteristic of low elevation. Values given are mean + SE a) effective quantum yield, b) stomatal conductance, and c) days to first flower. P-values in boldface remain significant afler sequential Bonferroni correction. Hybrid means not sharing letters differ significantly based on Tukey-Kramer adjusted comparison of least square means. 125 Aboveground biomass (g) o—: 000 Proportion survival 5: o low control high .0953 NAON _ c x2=2.31, df=2 P0.3155 k. i .. . low control high Hybrid selection regime Stem length (cm) Flower number 60‘ 40- 20- 4o- F2_98=1 .28 P=0.2823 low control high F2_98=0.19 P=0.8237 low control high Hybrid selection regime Figure 19. The effect of selection regime (low elevation, greenhouse control, or high elevation) on fitness components of hybrids grown in a temperature regime characteristic of low elevation. Values given are mean + SE a) aboveground biomass, b) stem length, c) proportion survival, and d) flower number. P-values in boldface remain significant after sequential Bonferroni correction. Hybrid means not sharing letters differ significantly based on Tukey-Kramer adjusted comparison of least square means. quantum yield, a chlorophyll fluorescence parameter indicating the photochemical efficiency of light energy use, and stomatal conductance, a gas exchange parameter indicating stomatal openness to water vapor and carbon dioxide, were both retained. The low elevation selected population of hybrids had a significantly higher effective quantum yield in low elevation temperatures than both the greenhouse control population and high elevation selected population (Table 18, Figure 18a). Selection regime also affected stomatal conductance, although difference was not significant after Bonferroni adjustment (Table 18, Figure 18b). Hybrid populations did not differ in the onset of flowering (Table 18, Figure 18c). The high elevation selected population attained significantly less aboveground biomass than either the greenhouse control or the low elevation selected population (Table 18, Figure 19a). Hybrids did not differ significantly in stem length, flower number, or survival (Table 18, Figure 19b, c, d). Selected and control hybrids did not differ in the measured leaf physiological traits when grown in high elevation temperatures (Table 19, Figure 20a, b). The high elevation selected population showed less tissue damage after first exposure to freezing temperatures than the low elevation selected population, although this difference did not remain significant afier Bonferroni adjustment (Table 19, Figure 20c, (1). Within the high elevation temperature regime, hybrids did not differ in the fitness components of growth and survival (Table 19, Figure 21). DISCUSSION Selection on trails In this study, I found evidence that selection favors early flowering at high elevation and increased leaf physiological capacity in hot temperatures at low elevation. 127 .3 A F2’83=0.19 8 0.25 B F2’83=1.68 >. 0.6 _ P=0.8259 gf; P=0.1922 5%. $08 0.20 a? 04 g 0.. 0.15 3 .— 2 Q g g 0.10 « g 0-2 s s 33 a 0.05 0.0 ,. 0.00- _ _. low control high low control high 60 40 50 c a F,_,5=3.93 D 122,850.57 3;, g, 30 _ P=0.5652 s c: 40 ‘ g a 'O B 'U as]; m 0:: a a: E V 20 - i': v o\° °\° 10 - 0 _ . low control high low control high Hybrid selection regime Hybrid selection regime Figure 20. The effect of selection regime (low elevation, greenhouse control, or high elevation) on two leaf physiological traits and post-freeze tissue damage of hybrids grown in a temperature regime characteristic of high elevation. Values given are mean + SE a) effective quantum yield, b) stomatal conductance, c) % tissue damage following freeze 1, and d) % tissue damage following freeze 2. The effect of selection regime on tissue damage following freeze 1 does not remain significant after sequential Bonferroni correction. 128 3° 14 l A F293:0-81 B 1:294:0-95 § 12 ~ p=0_4467 E 30 — P=0.3905 g 10 - e, i 8 « in 20 § 6 - .2 3, a _ D _ g 4 a 10 .<8 2 - 0 - 0 — " ~ low control high low control high Hybrid selection regime Hybrid selection regime c x2=0.64,df=2 a 1.0 - P=0.6407 . .?.. E 0.8 - E 0.6 - E Q 0.4 _ 8 ‘1‘ 0.2 - 0.0 low control high Hybrid selection regime Figure 21. The effect of selection regime (low elevation, greenhouse control, or high elevation) on fitness components of hybrids grown in a temperature regime characteristic of high elevation. Values given are mean + SE a) aboveground biomass, b) stem length, and c) proportion survival. 129 When significant patterns of selection were observed, selection always acted in the direction of the native parental species’ trait value, supporting the hypothesis that M. cardinalis photosynthetic traits and M. Iewisii flowering phenology are adaptive at their respective low and high elevation range centers. Despite clear evidence that selection favored early flowering at high elevation, I did not find differences in the onset of flowering among hybrid populations grown in a low elevation temperature regime. One possible explanation for this discrepancy is that flowering time has low heritability. However, this explanation seems unlikely because M. Iewisii and M. cardinalis display genetic differentiation for flowering time which should be segregating in interspecific hybrid populations. A more likely explanation is that I could not observe flowering phenology in a high elevation temperature regime, where differences in the onset of flowering may have been more pronounced than in the low elevation temperature regime. Differences in flowering phenology between parental species are much greater at high elevation than at low. It remains possible that selected hybrid populations will display differences in the onset of flowering when grown in a high elevation environment. In this study, such differences could not be analyzed because most plants did not flower within the high elevation temperature regime. Within the low elevation temperature regime, interspecific hybrid populations selected at low elevation displayed increased leaf physiological capacity relative to both an unselected control and to a population selected at high elevation. I did not detect selection on leaf physiological traits at high elevation, nor did the low elevation population experience a physiological cost when grown in a high elevation temperature regime. Photosynthetic acclimation to temperature is well documented (Billings et al. 130 1971; Berry and Bj6rkman 1980), and to understand the evolution of photosynthetic traits requires characterization of photosynthetic physiology across a range of relevant environments, so that population or genotypic physiological breadth as well as mean values are quantified. Other studies that have used segregating hybrid populations to measure natural selection on leaf morphology (e.g., Jordan 1991; Nagy 1997) and leaf physiology (e. g., Lexer et al. 2003; Ludwig et al. 2004) have also found selection operating in the direction of mean trait values in native populations or species. These studies have all used multivariate regression analysis to quantify within-generation phenotypic selection differentials and gradients on traits. In the present study, phenotypic selection differentials and gradients on physiological traits were not calculated. Instead, selection was evaluated as the difference in trait mean value between control and selected populations, an assessment of response to selection that could arise from direct phenotypic selection as well as underlying genetic correlations. Future studies combining within-generation multivariate selection analysis with measurement of between-generation selection responses will yield valuable information about the strength and direction of phenotypic selection, relationships among measured traits, and the trajectory of trait evolution. Between-environment fitness trade-offs A strength of the experimental evolution approach used here is the ability to examine not only patterns of trait evolution but also the fitness consequences of trait changes. If adaptation to one environment entails a cost to adaptation in another environment, then selected hybrid populations should display reduced fitness, relative to 131 an unselected control population, when grown in an environment in which they were not selected. One such tradeoff was observed in this study, where hybrids selected at high elevation displayed reduced biomass when grown in temperatures characteristic of low elevation. Other patterns of difference in fitness components between selected and control hybrid populations were suggestive of evolution of greater fitness within the selected environment at a cost to fitness within the unselected environment (e. g., survival of both selected populations was numerically higher than the control in the selected environment and lower than the control in the unselected environment; Figures 19c, 21c). However, no other fitness components exhibited significant between- environment tradeoffs, and in no case did measured fitness components significantly increase for populations grown in the environment in which they were selected, calling into question the effectiveness of selection and/or of the measurement conditions. Low ability to detect differences in fitness among hybrid populations may be due to several factors. First, populations have only experienced one generation of evolution in each environment, perhaps leaving considerable segregating variation within each population. Second, selected and unselected environments were simulated in grth chambers. The measurement of fitness components within growth chambers is not ideal for several reasons, including small sample size, poor flowering within the high elevation temperature regime, and the inability to simulate overwinter conditions. The latter two limitations apply to the high elevation temperature regime in particular, in which expected differences between the parental species were not detected. Although the lack of significant interspecific differences for some traits may be due to low power (e. g., post-freeze tissue damage), other traits displayed very small differences that 132 cannot be attributed to lack of power alone, suggesting that measurement conditions were not sufficiently favorable for M. Iewisii and high elevation selected hybrids. More definitive tests of the costs of adaptation to each environment will come from continued generations of experimental evolution and the reciprocal transplantation of selected populations to low and high elevation for a more thorough assessment of fitness. A related approach that will yield further insight into the causes and consequences of adaptation to alternate environments will combine the identification of quantitative genetic loci underlying traits of interest with field studies of their ecological effects. Segregating hybrid populations transplanted to low and high elevation can be used to identify quantitative trait loci (QTL) for fitness in each environment. The effects of major QTL can then be assessed with near-isogenic lines (NIL), containing single QTL regions from one species introgressed by repeated backcrossing into the genetic background of another. In this manner the phenotypic effects and fitness consequences of changes in single genomic regions can be characterized in environments within and beyond the species’ range, leading to greater understanding of evolutionary constraints on range expansion. 133 APPENDIX A Population sampling 134 cad .3. mmm 3 $3363 3 .mmmfim .3 0:83:33. 38.3 0330332 00.2.. 003 .95 3830330332 .:.3. 0 and 3 .3 .Q 82 83.033 .23.? .3 0:83:23. 383 .m 002 003 .309 :0380 .:.3. 3 end 3 .3 .Q 32 83.033 .mSSm .3 0:83:33. 383 .m :05. 0m: .303 :03300 .:.3. 0 3.3.3 .3. 32 @3333 .30.?” .30 0:30 388.3 .02 0 36 H RS 33.033 .35.? .30 0330:0333: 30333, 033E0mo> .32 3 333.33 .3. 3.3.2 333333 .3me .30 0330:0333: .30—30> 033808> .32 0 333.33 .3. 33.333 mood: .mmhfim .30 30:33:0m 303333 0338003; .32 3 omd .3. 3 3N3 mood: .95.? .30 30:33:0m 30333» 0338003; .32 0 end 3 .3 ..r o3m3 mend: .95.? .30 93:03. 3333/ 0338003; .32 3 mod 3 .3 ..3. 3333 News: .952” .30 98:03. 30330> 033E0mo> .32 0 N333 Q wom3 303.033 .36.? .3 30003032 3.83 .m 0:030? .32 3 VON D wom3 3.333 3 .32...” .3 300.3032 38.3 .m 0:02:35 .32 .0 333.33 ...3. 3383 N263 3 .mmufim .30 03000000 30330> 033E0mo> .32 3 mmd 3 ..3. oma 533.033 .53.?“ .30 38m .303 03:33.3. .32 0 03.33 .3. :3 32.33 3 .vovsm .3 3333303230 3.8.3 .3 3003 8:030:03 .32 0 333.33 ...3 .3 ..3. cow @8333 .8me .30 3:03 3:33 3:350 MW033330332 .32 0 and Q mmw 3 5.03 3%.? .3 05:33.3. 38.3 .m 3003 3033:3303 5.3. 0 3nd 3 .0 0mm 3%on 3.3.55” .30 0:832 030300032 30:3 .:.3. 0 omd .3. can 3.33.83 33.3.... .3 30230 33.3 «>03 0:333. .:.3. 0 $3 .3. moo 0:333 3.3.3 .30 0:80 33.53 33 .32 0 In 3 .3 com @3333 .nwvfim .30 08“33.32 00933332 .32 0 end 3. gm 033.033 .2.ch .30 0832 33:03 33 .32 0 cad .3. mum mmod: .9.on .30 :030E003m 33.53 33 .32 0 cod .3. mum 26.033 .9092” .3 3003032 33.83 33 .32 0 333203 033.» .3. $033 >2. .2. 08:00:03? 33:30:03 5:380 m03o0n3m 0:83:33. 0 33.3. 08333332 0 32 00:80 3.3.3.3303 .32 n 3 53:33:80 .333 n 0 “0030035 ”030338 00 €0c3>053< .333 3:03.898 8:23:03 :0 .333 3533305.: 30083300: 3.3.3 00:30:30 0030 03% .30 $000.53 .53 33080 0333338308030 3003333030 .83 5830 803332323 .30 $8 30¢ 03330333033 b33030 3:333 30:0 .383 53333030 .33. .730 0030:3280 03033303 30:0 0303333 608830303 3303:3053 30:00: 5:33:03 5330003 3:000:33 0030035 .33.. 03330 .3. 135 3533033 330:6 w:3.:6 8:330 83332303... as 3 .3 d 88 8g: :3? .3 05:5 8.3 3.5.3 553 a: 3 a; _ .3 d 88 3;: 3.3.2 .30 38m 33 3.2 .fi 3 I 3 an” 32: .20.: ..8. 8.855 08: .023. .3. 3 a: .3 22 $3.: .30.: .30 2.52 833 328. .fi 3 :3 .3 :E ~35: .30.? .30 39856 33 8002 a: 3 Rd 3 .3 2mm :2: .02.: .30 05:38.: E: 2.33.5.3 ...2 3 w; 3 was 23: «3.2 .30 33.5.5 82 28: 3020 33585 a: 3 So 3 3.3 $2: .NNEM .30 356% 2.3: 325an .22 3 ”to .3 $8 9.2: 35.2 .30 38m .333 38.0 38m a: 3 80 3 .3 .3. 8.3 $2: .55” .30 383.53. a: 30225 a: 3 32 3 SM: 83: .33.? .30 8830 3.3 3855.3 ..2 3 3o 3 Ne: 62: .23.? .30 2.82:: as: .8883 a: 0 N3 3 82 :32 .3333 .3 3.320 :3 3 So .3. 3.: :83: $2... .30 832520 2233 a: 3 m2 3. 3: 83m: .08.”... .3 05.5383 35.3 .2 580:3 .5 0 So 3 #2 $3.23 33.3 .3 85283 38.3 2022 82 a: 3‘3 3832022 .3. 3 30.380 2 2.3 136 APPENDIX B Calculation of seed dormancy and recruitment parameters 137 Calculation of seed dormancy parameters, following the methods of Horvitz and Schemske (1995) Mimulus Iewisii. May Lalge (rage center. 2690 m) a) Germination percentage from seed = 2291 seedlingszoo3/21,6OO seedszooz = 10.6% seedlings/seed b) Number of dormant seeds in 2003 = 63 seedlingszom/OJO6 seedlings/seed = 594 seeds c) Percentage of dormant seeds in 2003 = 594 dormant seedszoo3/21,6OO seedszooz = 2.8% (1) Percentage survival of seeds in 2003 = percentage germinated + percentage dormant = 10.6% + 2.8% = 13.4% Mimulus cardinalis, BM Meadows (range center. 830 m) a) Germination percentage from seed = 38 seedlings in 2003/2,500 seeds in 2002 = 1.5% seedlings/seed b) Dormant seeds in 2003 = 7 seedlings in 2004/0.015 seedlings /seed = 460 seeds 138 c) Percent dormancy in 2003 = 460 dormant seeds/2500 seeds in 2002 = 18.4% d) Percent survival of seeds from 2002 to 2003 = 1.5% + 18.4% = 19.9% Calculation of recruitment parameters Estimates of seed germination from the seed stations could not be used for matrix calculations because the rate of seed germination in the stations far exceeded the rate of seedling recruitment. In other words, seedling germination observed in the seed stations was truncated before plants became fully established, causing the rate of germination to overestimate the rate of juvenile recruitment. Better estimates of transitions from seed to vegetative classes for each site and year were obtained by calculating the ratio of recruitment to seed production. Because seeds are known to live at least one year in the seed bank, seed production was estimated as the moving sum across a two-year window, yielding the following estimate of transitions from seed to a particular stage class: an = number of recruits in class l,+1/(S€Cd production, + seed production“) Because the dormant seed bank is of unknown size and age, this calculation may underestimate the true denominator. To assess the affects of uncertainty in estimates of recruitment on population growth, I increased the size of the seed bank by as much as 50% and found that lambda decreased by 1.4 — 6.7% for M. cardinalis and 0.7 — 5.7% for M. Iewisii. Lambdas at all locations responded similarly to increases or decreases in 139 the size of the seed bank, and the magnitude of change in lambda due to variation in seed bank size was not sufficient to erase differences in lambdas between central and marginal populations. 140 APPENDIX C Transition matrices, sensitivity matrices, and lambdas 141 ace—con cod and Ed 3&3: $3.9 vowed o o .8un vmd mmd 86 3.35 ooomd meemd momod mommd .E .wq 833 - 838 So So cod 3.9. o o o 38¢ .3 .am 3.36 cod cod cod mod 0 _ a 3; o 9% _ .0 team «53-33 Nvd who mod 3.33 vwomd ammo 362. o .oaom 3.0 :3 mod 3.32 936 ammmd amnmd m-oE.~ .8 .mA Cmmwd - omomdv mmd end mod 3.3 2 o 386 $86 maid .3 .8m 9306 cod cod cod m _ .o 32 own cum ova _ .o poem «:34ch Mad we; hmd was 2: _ whhmd 336 o 5-3 m.» .oaom mod 36 Ed oflvvcm ooomd 392. o mbvfim .E .wq ammo; .. mmovdv cod cod cod cod o o o wooed .8 .8m wowed cod cod cod and $8 ”mom o wand Bow ~3NA5=~ AB 38 ate—.32 :25 535.83 g GU .xcmov w 650% ..E .wq .8 .8m comm 93% .8 .mA .3 .8m Bow x52: bmzzmcom 5.92: 828.95. dig some 323 momoficoaa E cozw Coca $388 a com flmtofim oocowaaoo $3 cocoobooéfim 638:852 n ...oaoM: was A50 m A fiwcfl 8on 5.333 .3180 m A 5wc2 8on ”332.859 $5 0385858-:0: 033 u ...E .qu A80 m w Swee— Eocm ”53$ .3 ”80 m w ~3qu 8on 53533 «.5 o>uo=woa2éoc :25 n ...E .Em: ”mic—3m 8 8233059" mam—o owfim .EEBE 52285 380» Ea £0382 .86on :08 com w 98 £25; 33:28 .8392: gamma—EH ._ U 2an 142 Ed Rd add mddmm 3‘de :dmd waded médmd .E .wq Gomod - mdwvdd vod nod Nod Q62 Rodd mvmdd mvmfid moved .5 .8m 33d odd odd odd dd 83 mg m 2d doom ..9—ecu dwd mod em; mcdmdm mmmvd dde d d .oaom mod :d odd mud? wmmmd ova—d aged #10de .5 .mA 9: Ed - mwmmdv odd 8d 5d de— d didd midd v-03; .E .Em vmhvd ddd odd odd wdd dz: mm d 032d doom 83.83 mod vwd mfid dddmmm Ngvd 32d wmddd 5-3:, .oaom mmd Rd cod 1 dew mommd dommd $N~d mode; .8 .wq Avomnd - wdmmdd odd wdd 8d $.va d Rodd NE _d médmd .8 .8m ammod ddd ddd odd mod ohm“ dwm w ded doom Nddnéddm mvd vmg mvd M368— w E mvd ddmdd d 0.00 _ A .oaom 2d 2d Cd 3.23 85d mEvd wdddd 905m; .8 .mA €30d - onwvdd mdd 3d mod 3N: Ebdd muddd oddmd mocha .5 .8m dmhmd ddd odd ddd mod mod 2. d med doom Sdudddn A8 mmmv .2:— Bonfiam £33.53.» .2 mmd Sd Ed 3: 5v Emvmd _d_ _d Svod 5-08; 6&3“ n _ d Ed wdd m E .mem mmhmd Sawmd dmmfid mbdmd .E .wq Chad - cwmmdd odd Ed mod Nddmw d good dfldd m-omm.m .2 .8m awed ddd odd ddd dmd 33 meg :3 2d doom .3288 _U 033. 143 m m d dcd on. E Evdccdo mcdbd mmm E d E Nod Dodo. E 683E Ed Ed 3d nwdEVdE meEd mmmmd Eumddd Eoddd .8 .wE AmdmmE - wedwdv vod vod add EE.E.mEv d medd EdEd mbmdd .8 .8m Emdd odd odd ddd Ed $de NEE don dmd doom Eve—:2. mud end mmd modvcmm ond mNbdd EdEdd d .083E Ed dEd m Ed Vm. EEh mmdmd wohmd 2: Ed EoweE .8 .wE Amde - 9&de mod mod mod wm.3mE d wmmdd hmvEd m-odE.E .8 .8m good ddd ddd odd mod can: an: Evm dvEd doom 83-83 wvd mm d 3. E «mam «do mmmhd mcmmd vadd Womw. E 680% Ed dmd omd 2.3 Ecm wvad mwvmd dded oéood .8 .wE Emdva - mmEodv mod vdd 3.0 Ede. d EEmdd Ndhdd m-oEd.N .8 .8m wmhd. E ddd ddd odd de SENS mmwom Ehm cmdmd doom Nddnéddn mvd cvd EE.E dodmwmw EEwd mummd Nvmdd Woddd .oaom Ed Ed add 8.83m mdwdd mmcmd ENSd 0-33” .8 .wE AmdmoE - ENMdEV mod mdd Ed 3de d mde dmmdd Womvd .8 .8m wmvm. E odd ddd ddd cmd munmmE Emdwm Eon m E mmd doom 33.33 A8 mdNEV «835$ £33889 .3 3d E E. E mmd mvdEm mwvvd wmmdd m Eddd 93:: 683E $3803 EU 2an 144 wt; 8.8 8.8 8.8 8.0 88.8 o o 88.8 88 88-88 E... 83 22:3 .3§ .2 88 8.8 8: 8.8.8: 888 8:8 $8.8 5-3.3. .25 2.8 8.8 3.8 8.88 8:8 82.8 :88 38; .a .3 £87338 8.8 8.8 8.8 8.88 8 2.8.8 88.8 388 .aém 8...: 8.8 8.8 8.8 8.8 888 88 8» So 8% Ego—:2. 8.8 n: 8.8 $888 ~88 82.8 :88 39% .288 8.8 8.8 2.8 2.88 9.2.8 E88 88.8 388 .a .3 682. 38.8 :8 8.8 :8 8.8 o o 88.8 388 8.5 28.8 8.8 88 88 8° 8% 8mm 8 28.8 88 88-88 8.8 So of 8.85. 3.88 $8.8 88.8 o .288 :8 8.8 8.8 8882 88.8 52 8:8 8-09.8 .8 .3 982.7838 8.8 8.8 2.8 8.:8 o 3.88 28.8 92.2 .228 88: 8.8 8.8 8.8 2.8 GE 58 82 82.8 88 838a $8 8.8 88 8.5% N83 N82 :88 3:8 .888 8.8 88 Ed 8.5:: 82.8 :98 28.8 8-088 .2 .3 38.7289 88 8.8 8.8 8.82 o 88.8 88.0 ERA 52% $8; 8.8 8.8 88 2.8 88:. 83; 88 8.8.8 88 $8.88 a: 8m: 8:3 .8233 .8 .383 6 289 145 88 88.8 $8 $888 8888 8:8 88.8 8.888 .2838 2.8 88.8 88.8 88.288 88.8.8.8 828.8 58.8 8-88.8 .a .3 52.78888 88 :8 88 88.88: 8 888.8 888.8 8-083. .8388 888.8 88.8 88.8 88.8 :8 8882 8:88 8: $88 888 88-888 As 88: 82.30 .8323 .8 88.8 $8 8.8 8.88:: 888.8 82.8 8 8 .8888 $8 888 28 88:2 88:8 888 8 8.82 .a .3 8888.848888 88.8 88.8 88.8 88.8 8 8 88.8 8-88.8 .8388 8888.8 88.8 88.8 88.8 8.8 88888 88 8 8.8 888 3.2:“ 88.8 8.— 88.8 88.8 888.8 82.8 8 8 .2898 88.8 8.8 88.8 88.8 8888 88.8 8 8-882 .2 .3 @888 -2888 88.8 88.8 88.8 88.8 8 8 8888 8-388 8.88 888.8 88.8 88.8 88.8 88.8 888 N8: 8 888.8 888 8883888 $8 888 88.8 8.882 88.8 8888.8 8 8 .2888 8.8 :8 :8 8.88: 8838 $888 8 8.88.8. .a .3 8888-89.88 88.8 88.8 88.8 88.8 8 8 8 8-88.8 8.88 8888.8 88.8 88.8 88.8 88 82 8 88 8 888.8 8888 N838” 8._ 88 88 88828 8:88 8 8 8 .2838 88.8 88.8 88.8 88.8 888 88.8 8 8.388 ..a .3 $388459 88.8 88.8 88.8 88.8 8 8 8 8-882 8.88 8.8888 5 2888 146 888.8 88.8 88.8 88.8 888 :88 888: 8:. 8888 8888 888-888 888 8.88 88 88.88888 8888 88.8 8 88-883 88838 8.8 88.8 88.8 888.8883. 888.8 888.8 888 8.88.8 .888 .3 8888.8-888.88 888 88.8 88.8 8.888 8 88.8 $888 8-88:8 888.8888 888.8 88.8 88.8 88.8 88.8 8888 8888.8 8 8888 8888 888-888 8:. 88888 383 882 8.82.8 .88 88.8 88.8 88 88.888 883.8 88888 8888 8.88.8 .2838 88.8 28 88.8 88888 888.8 8 8 88.8 8888 8-8888 .8: .3 888.888.38.88 8.8 8.8 888 8.88 8 8888.8 8888 8.288 888.88 888.8 88.8 88.8 88.8 :8 888888 888 88 8.8.8 8888 .8282— 888 3.8 88 8.8838 8E8 888 8888 E88 .8898 88.8 88.8 28 88.888 8888 888.8 888.8 8-88.8 .888 .3 8888.8 - 88888 88.8 88.8 88.8 88.88 8 8888 888 8.88.8 8.88 888.8 88.8 88.8 88.8 88.8 88 8 m 88.8 88 8 88.8 8838 88388 88 88.8 888 88.888 8888 888.8 8 8 .2838 88.8 88.8 88 8888.88. 828 888.8 8888.8 8-88.8 .888 .3 8888-88.88 88.8 88 88.8 8.888 8 8 8288 8.08.88 8.88 888.8 88.8 88.8 88.8 8.8 88888 888 8 888.8 8888 88.88 88.88888 80 2888 147 88.8 88.8 88.8 88.888888 88888 8.8888 888.8 8-88.8 .2838 88.8 888 88.8 8.8888 8888.8 8888.8 88888 8.88888 .88 .3 8888888888.: 888 88.8 :8 88.8888 8 88888 88888 8-88.8 88.88 8888.8 88.8 88.8 88.8 88.8 888888 88888 888 8888.8 888 888-888 88.8 888 8:8 88.88: 8888 8888.8 8 8 .8838 88.8 88.8 88.8 88.888888 88888 88888 8 8-8888. .8 .3 88888888888 88.8 88.8 88.8 88.8 8 8888.8 888.8 8888.8 8.8888 88.8.8.8 88.8 88.8 88.8 888 8888 N888 8 888.8 8888 8888-888 8:. 88888 ...8... 882.883 .8338 .88 88.8 88.8 88.8 88.888888 8888.8 8888.8 888.8 8-88.8 .2838 $8 888 8.88 88.8828 888 8888.8 8888 8-88: .888 .3 85.88.888.88 88 88 88.8 888888 8 8888.8 888.8 8.888 888.8888 888.8 88.8 88.8 88.8 88.8 8888 88888 888 888 8888 @223 88.8 88.8 8.88 888888 88388 888.8 8 8.88.8 .2838 888 88.8 888 88.88888 8888 8888.8 8888.8 8-88.8 .888 .3 6888848888 888 88.8 888 R888 8 888 8888.8 8.8888. 88.88 8888.8 88.8 88.8 88.8 88.8 88888 8888 8 888.8 8888 888.888 88.8 88.8 $8 88.88888 8888 888.8 88888 8 .2838 88.8 88.8 88 8.88888 8888.8 8888.8 8888.8 8-8888 .8 .3 822-8888 88.8 88.8 88.8 88.388 8 8 8888 8.888 888.8888 888.8888 6 088888 148 88888.8 - 888.88 RSA A338 - @88de 886 3d m _ .o mod cod mmd :6 mod cod 36 3d mod cod 5.— de mod cod ma; 36 Q .o cod NNN end nod cod No.36; 38.9%: 3.2 3 Ed Edmomm wvdnmg av. 38: mod 88.88888 80 88.888 ..8—2:— mccnéecm vmvwd omfid :03 mind cwmfio wcmwh 326 :VNod 9186 3% _ mowed good omwod owmw cvood men 8 .o cvwfio wmm o cmood m3 :8 o wéaod cémmm 938m mod o 5-00.8.5 m-om©.m owmod .0538 ..8: .mA .8 .8m 8888 died .8: .wq ..8 .8m 8888 149 LITERATURE CITED Anthony, K. R. N., and S. R. Connolly. 2004. Environmental limits to growth: physiological niche boundaries of corals along turbidity-light gradients. Oecologia 141:373-384. Antonovics, J. 1976. The nature of limits to natural selection. Ann. Mo. Bot. Gard. 63:224-247. Amtz, A. M., and L. F. Delph. 2001. Pattern and process: evidence for the evolution of photosynthetic traits in natural populations. Oecologia 127:455-467. Amtz, A. M., E. H. DeLucia, and N. Jordan. 1998. Contribution of photosynthetic rate to growth and reproduction in Amaranthus hybridus. Oecologia 117:323-330. Arrontes, J. 1993. Nature of the distributional boundary of F ucus serratus on the north shore of Spain. Mar. Ecol. Prog. Ser. 93:183-193. Asselin, H., S. Payette, M. J. Fortin, and S. Vallee. 2003. The northern limit of Pinus banksiana Lamb. in Canada: explaining the difference between the eastern and western distributions. J. Biogeogr. 30:1709-1718. Battaglia, M., C. L. Beadle, and S. Loughhead. 1996. Photosynthetic temperature responses of Eucalyptus globulus and Eucalyptus nitens. Tree Physiol. 16:81-89. Beardsley, P. M., A. Yen, and R. G. Olmstead. 2003. AP LP phylogeny of Mimulus section Erythranthe and the evolution of hummingbird pollination. Evolution 57:1397-1410. Berry, J ., and O. Bjorkman. 1980. Photosynthetic response and adaptation to temperature in higher plants. Annual Review of Plant Physiology 31 1491-543. Billings, W. D., P. J. Godfrey, and D. P. Bourque. 1971. Metabolic acclimation to temperature in arctic and alpine ecotypes of Oxyria digyna. Arct. Alp. Res. 3:277-289. 150 Blackburn, T. M., K. J. Gaston, R. M. Quinn, and R. D. Gregory. 1999. Do local abundances of British birds change with proximity to range edge? J. Biogeogr. 26:493-505. Bolhar-Nordenkampf, H. R., and G. unist. 1993. Chlorophyll fluorescence as a tool in photosynthesis research. Pp. 193-206 in D. 0. Hall, J. M. O. Scurlock, H. R. Bolhar-Nordenkampf, R. C. Leegood and S. P. Long, eds. Photosynthesis and Production in a Changing Environment: A Field and Laboratory Manual. Chapman and Hall, London. Bongi, G., and S. P. Long. 1987. Light dependent damage to photosynthesis in olive leaves during chilling and high temperature stress. Plant Cell Environ. 10:241- 249. Bradshaw, A. D. 1960. Population differentiation in Agrostis tenuis Sibth. III. Populations in varied environments. New Phytol. 59:92-103. Bradshaw, A. D., and T. McNeilly. 1991. Evolutionary response to global climatic change. Ann. Bot. 67:5-14. Bradshaw, H. D., K. G. Otto, B. E. Frewen, J. K. McKay, and D. W. Schemske. 1998. Quantitative trait loci affecting differences in floral morphology between two species of monkeyflower (Mimulus). Genetics 149:367-382. Bradshaw, H. D., and D. W. Schemske. 2003. Allele substitution at a flower colour locus produces a pollinator shift in monkeyflowers. Nature 426: 176-178. Brown, J. H. 1984. On the relationship between abundance and distribution of species. Am. Nat. 124:255-279. Brown, J. H., D. W. Mehlman, and G. C. Stevens. 1995. Spatial variation in abundance. Ecology 76:2028-2043. Brown, J. H., G. C. Stevens, and D. M. Kaufman. 1996. The geographic range: size, shape, boundaries, and internal structure. Annu. Rev. Ecol. Syst. 27:597-623. Brys, R., H. Jacquemyn, P. Endels, G. De Blust, and M. Hermy. 2004. The effects of grassland management on plant performance and demography in the perennial herb Primula veris. J. Appl. Ecol. 41 :1080-1091. 151 Bullock, J. M., R. J. Edwards, P. D. Carey, and R. J. Rose. 2000. Geographical separation of two Ulex species at three spatial scales: does competition limit species' ranges? Ecography 23:257-271. Carbonell, R., and J. L. Telleria. 1998. Increased asymmetry of tarsus-length in three populations of Blackcaps Sylvia atricapilla as related to proximity to range boundary. Ibis 140:331-333. Carter, R. N., and S. D. Prince. 1981. Epidemic models used to explain biogeographical distribution limits. Nature 293:644-645. . 1988. Distribution limits from a demographic viewpoint. Pp. 165-184 in A. J. H. Davy, M. J .; Watkinson, A. R., ed. Plant population ecology. Blackwell Scientific Publications, Oxford. Caswell, H. 2001. Matrix Population Models: Construction, Analysis, and Interpretation. Sinauer Associates, Sunderland, MA. Charlesworth, B. 1980. Evolution in age structured populations. Cambridge University Press, Cambridge, MA. Clausen, J. C., D. D. Keck, and W. M. Hiesey. 1940. Experimental studies on the nature of species. I. Effect of varied environments on western North American plants. Carnegie Institute of Washington publ. no. 520. . 1948. Experimental studies on the nature of species. III. Environment responses of climatic races of Achillea. Carnegie Institution of Washington Publication 581, Washington, D. C. Cleavitt, N. 2004. Comparative ecology of a lowland and a subalpine species of Mnium in the northern Rocky Mountains. Plant Ecol. 174:205-216. Cline, M. G., and A. O. Agatep. 1970. Temperature and photoperiodic control of developmental responses in climatic races of Mimulus. Plant Cell Physiol. 1 1:609-619. Close, D. C., and C. L. Beadle. 2003. Chilling-dependent photoinhibition, nutrition and growth analysis of Eucalyptus nitens seedlings during establishment. Tree Physiol. 23 :2 1 7-226. 152 Cossins, A. R., and K. Bowler. 1987. Temperature Biology of Animals. Chapman and Hall, London. Criddle, R. S., M. S. Hopkin, E. D. McArthur, and L. D. Hansen. 1994. Plant distribution and the temperature coefficient of metabolism. Plant Cell Environ. 17:233-243. Crozier, L. G. 2004. Field transplants reveal summer constraints on a butterfly range expansion. Oecologia 141:148-157. Cumming, G. S. 2002. Comparing climate and vegetation as limiting factors for species ranges of African ticks. Ecology 83:255-268. Cunningham, S. C., and J. Read. 2002. Comparison of temperate and tropical rainforest tree species: photosynthetic responses to grth temperature. Oecologia 133:112-119. . 2003. Do temperate rainforest trees have a greater ability to acclimate to changing temperatures than tropical rainforest trees? New Phytol. 157:55-64. Curnutt, J. L., S. L. Pimm, and B. A. Maurer. 1996. Population variability of sparrows in space and time. Oikos 76:131-144. Dahl, E. 1951. On the relation between summer temperature and the distribution of alpine vascular plants in the lowlands of Fennoscandia. Oikos 3:22-52. Davison, A. W. 1977. The ecology of Hordeum murinum L.: III. Some effects of adverse climate. J. Ecol. 65:523-530. Denham, A. J ., and T. D. Auld. 2004. Survival and recruitment of seedlings and suckers of trees and shrubs of the Australian arid zone following habitat management and the outbreak of Rabbit Calicivirus Disease (RCD). Austral. Ecol. 29:585- 599. Doherty, P. F ., T. Boulinier, and J. D. Nichols. 2003. Local extinction and turnover rates at the edge and interior of species' ranges. Ann. 2001. F enn. 40: 145-1 53. 153 Dorken, M. E., and C. G. Eckert. 2001. Severely reduced sexual reproduction in northern populations of a clonal plant, Decodon verticillatus (Lythraceae). J. Ecol. 89:339-350. Dudycha, J. L., and A. J. Tessier. 1999. Natural genetic variation of life span, reproduction, and juvenile growth in Daphnia. Evolution 53: 1744-1756. Edgington, E. S. 1995. Randomization tests. Marcel Dekker, Inc., New York. Endler, J. A. 1986. Natural selection in the wild. Princeton University Press, Princeton, N. J. Falk, S., D. P. Maxwell, D. E. Laudenbach, and N. P. A. Huner. 1996. Photosynthetic adjustment to temperature. Pp. 367-385 in N. R. Baker, ed. Photosynthesis and the Environment. Kluwer, Dordrecht. Faugeron, S., E. A. Martinez, J. A. Correa, L. Cardenas, C. Destombe, and M. Valero. 2004. Reduced genetic diversity and increased population differentiation in peripheral and overharvested populations of Gigartina skottsbergii (Rhodophyta, Gigartinales) in southern Chile. J. Phycol. 40:454-462. Ferris, R., I. Nijs, T. Behaeghe, and I. Impens. 1996. Elevated C02 and temperature have different effects on leaf anatomy of perennial ryegrass in spring and summer. Ann. Bot. 78:489-497. Fisher, R. A. 1930. The genetical theory of natural selection. Clarendon Press, Oxford. Flebbe, P. A. 1994. A regional view of the margin - Salmonid abundance and distribution in the Southern Appalachian Mountains of North Carolina and Virginia. Trans. Am. Fish. Soc. 123:657-667. Fox, G. A. 2001. Failure time analysis: studying times to events and rates at which events occur in S. M. Scheiner and J. Gurevitch, eds. Design and analysis of ecological experiments. Oxford Univ. Press, Oxford, UK. Fréville, H., B. Colas, M. Riba, H. Caswell, A. Mignot, E. Imbert, and I. Olivieri. 2004. Spatial and temporal demographic variability in the endemic plant species Centaurea corymbosa (Asteraceae). Ecology 85:694-703. 154 Garcia, D., R. Zamora, J. M. Gomez, P. Jordano, and J. A. Hodar. 2000. Geographical variation in seed production, predation and abortion in Juniperus communis throughout its range in Europe. J. Ecol. 88:436-446. Garcia-Ramos, G., and M. Kirkpatrick. 1997. Genetic models of adaptation and gene flow in peripheral populations. Evolution 51:21-28. Gaston, K. J. 1990. Patterns in the geographical ranges of species. Biol. Rev. 65:105- 129. . 2003. The structure and dynamics of geographic ranges. Oxford Univ. Press, Oxford, UK. Gilbert, N. 1980. Comparative dynamics of a single-host aphid. I. The evidence. J. Anim. Ecol. 49:351-369. Gomulkiewicz, R., R. D. Holt, and M. Barfield. 1999. The effects of density dependence and immigration on local adaptation and niche evolution in a black- hole sink environment. Theor. Popul. Biol. 55:283-296. Gonzalez-Guzman, L. I., and D. W. Mehlman. 2001. Developmental stability across the breeding distribution of the Scissor-Tailed Flycatcher ( T yrannus forficatus). Ecol. Lett. 4:444-452. Good, R. D. O. 1931. A theory of plant geography. New Phytol. 30: 149-171. Grace, J. 1987. Climatic tolerance and the distribution of plants. New Phytol. 106:113- 130. Grant, V. 1963. The origin of adaptations. Columbia Univ. Press, New York, NY. Graves, J. D., and K. Taylor. 1986. A comparative study of Geum rivale L. and G. urbanum L. to determine those factors controlling their altitudinal distribution. 1. Growth in controlled and natural environments. New Phytol. 104:681-691. . 1988. A comparative study of Geum rivale L. and G. urbanum L. to determine those factors controlling their altitudinal distribution 11. Photosynthesis and respiration. New Phytol. 108:297-304. 155 Griggs, R. F. 1914. Observations on the behavior of some species at the edges of their ranges. Bull. Torrey Bot. Club 41:25-49. Grinnell, J. 1917. Field tests of theories concerning distributional control. Am. Nat. 51:115-128. Guo, Q. E, M. Taper, M. Schoenberger, and J. Brandle. 2005. Spatial-temporal population dynamics across species range: from centre to margin. Oikos 108:47- 57. Haldane, J. B. S. 1956. The relation between density regulation and natural selection. Proc. R. Soc. Lond. B 145:306-308. Haldimann, P., and U. Feller. 2004. Inhibition of photosynthesis by high temperature in oak (Quercus pubescens L.) leaves grown under natural conditions closely correlates with a reversible heat-dependent reduction of the activation state of ribulose-1,5-bisphosphate carboxylase/oxygenase. Plant Cell Environ. 27:1169- 1183. Harvey, P. H., and M. D. Pagel. 1991. The comparative method in evolutionary biology. Oxford University Press, Oxford. Heller, H. G., and D. M. Gates. 1971. Altitudinal zonation of chipmunks (Eutamias): energy budgets. Ecology 52:424-433. Hengeveld, R., and J. Haeck. 1982. The distribution of abundance. 1. Measurements. J. Biogeogr. 9:303-316. Hennenberg, K. J ., and H. Bruelheide. 2003. Ecological investigations on the northern distribution range of Hippocrepis comosa L. in Germany. Plant Ecol. 166:167- 188. Hersteinsson, P., and D. W. Macdonald. 1992. Interspecific competition and the geographical distribution of Red and Arctic Foxes Vulpes wipes and Alopex lagopus. Oikos 64:505-515. Hickman, J. C. 1993. The Jepson manual: higher plants of California. Univ. of California Press, Berkeley, CA. 156 Hiesey, W. M., M. A. Nobs, and O. Bjorkman. 1971. Experimental studies on the nature of species. V. Biosystematics, genetics, and physiological ecology of the Erythranthe section of Mimulus. Carnegie Institute of Washington publ. no. 628. Hoffman, A. A., and M. W. Blows. 1994. Species borders: ecological and evolutionary perspectives. Trends Ecol. Evol. 9:223-227. Holt, R. D. 1996. Adaptive evolution in source-sink environments: direct and indirect effects of density-dependence on niche evolution. Oikos 75: 182-192. . 2003. On the evolutionary ecology of species' ranges. Evol. Ecol. Res. 5:159- 178. Holt, R. D., and M. S. Gaines. 1992. Analysis of adaptation in heterogeneous landscapes - implications for the evolution of fundamental niches. Evol. Ecol. 6:433-447. Holt, R. D., and T. H. Keitt. 2000. Alternative causes for range limits: A metapopulation perspective. Ecol. Lett. 3:41-47. . 2005. Species' borders: a unifying theme in ecology. Oikos 108:3-6. Holt, R. D., T. H. Keitt, M. A. Lewis, B. A. Maurer, and M. L. Taper. 2005 . Theoretical models of species' borders: single species approaches. Oikos 108:18-27. Horvitz, C. C., and D. W. Schemske. 1995. Spatiotemporal variation in demographic transitions of a tropical understory herb - Projection matrix analysis. Ecol. Monogr. 65: 155-192. Huberty, C. J ., and J. D. Morris. 1989. Multivariate analysis versus multiple univariate analyses. Psychol Bull 105:302-308. Huff, D. R., and L. Wu. 1992. Distribution and inheritance of inconstant sex forms in natural populations of dioecious buffalograss (Buchloe dactyloides). Am. J. Bot. 79:207-215. Hughes, L. 2000. Biological consequences of global warming: is the signal already apparent? Trends Ecol. Evol. 15:56—61. 157 Hummel, H., R. H. Bogaards, G. Bachelet, F. Caron, J. C. Sola, and C. Amiard-Triquet. 2000. The respiratory performance and survival of the bivalve Macoma balthica (L.) at the southern limit of its distribution area: a translocation experiment. J. Exp. Mar. Biol. Ecol. 251285-102. Huner, N. P. A., W. Migus, and M. Tollenaar. 1986. Leaf C02 exchange rates in winter rye grown at cold-hardening and nonhardening temperatures. Can. J. Plant Sci. 66:443-452. Huntley, B. 1991. How plants respond to climate change - migration rates, individualism and the consequences for plant communities. Ann. Bot. 67: 1 5-22. Hurry, V. M., and N. P. A. Huner. 1991. Low growth temperature affects a differential inhibition of photosynthesis in spring and winter wheat. Plant Physiol. 96:491- 497. Iio, A., H. Fukasawa, Y. Nose, and Y. Kakubari. 2004. Stomatal closure induced by high vapor pressure deficit limited midday photosynthesis at the canopy top of F agus crenata Blume on Naeba mountain in Japan. Trees Struct. Func. 18:510- 517. Ives, A. R., and E. D. Klopfer. 1997. Spatial variation in abundance created by stochastic temporal variation. Ecology 78: 1907-191 3. Jarvinen, A., and R. A. Vaisanen. 1984. Reproduction of Pied Flycatchers (F icedula hypoleuca) in good and bad breeding seasons in a northern marginal area. Auk 1 0 l :43 9-450. Jonas, C. S., and M. A. Geber. 1999. Variation among populations of Clarkia unguiculata (Onagraceae) along altitudinal and latitudinal gradients. Am. J. Bot. 86:333-343. Jones, R. H., and R. R. Sharitz. 1998. Survival and growth of woody plant seedlings in the understorey of floodplain forests in South Carolina. J. Ecol. 86:574-587. Jordan, N. 1991. Multivariate analysis of selection in experimental populations derived from hybridization of two ecotypes of the annual plant Diodia teres W. (Rubiaceae). Evolution 45 158 Juenger, T., and J. Bergelson. 2000. The evolution of compensation to herbivory in Scarlet Gilia, Ipomopsis aggregata: herbivore-imposed natural selection and the quantitative genetics of tolerance. Evolution 54:764-777. ' Jump, A. S., and F. I. Woodward. 2003. Seed production and population density decline approaching the range-edge of Cirsium species. New Phytol. 160:349-358. Kao, W. Y., T. T. Tsai, and W. H. Chen. 1998. A comparative study of Miscanthus floridulus (Labill) Warb and M. transmorrisonensis Hayata: photosynthetic gas exchange, leaf characteristics and growth in controlled environments. Ann. Bot. 81 :295-299. Kawecki, T. J. 1995. Demography of source-sink populations and the evolution of ecological niches. Evol. Ecol. 9:38-44. Keith, D. A. 2002. Population dynamics of an endangered heathland shrub, Epacris stuartii (Epacridaceae): recruitment, establishment and survival. Austral. Ecol. 27:67-76. Kimura, M. T. 2004. Cold and heat tolerance of drosophilid flies with reference to their latitudinal distributions. Oecologia 140:442-449. Kirkpatrick, M., and N. H. Barton. 1997. Evolution of a species range. Am. Nat. 150:1- 23. Krause, G. H., and E. Weis. 1984. Chlorophyll fluorescence as a tool in plant physiology. II. Interpretation of fluorescence signals. Photosynth. Res. 5: l 39- 157. . 1991. Chlorophyll fluorescence and photosynthesis: the basics. Annu. Rev. Plant Physiol. Plant Mol. Biol. 42:313-349. Kuehl, R. O. 2000. Design of experiments: statistical principles of research design and analysis. Duxbury Press, Pacific Grove, CA. Laisk, A., B. H. Rasulov, and F. Loreto. 1998. Therrnoinhibition of photosynthesis as analyzed by gas exchange and chlorophyll fluorescence. Russ. J. Plant Physiol. 45:412-421. 159 Larcher, W. 1995. Physiological Plant Ecology; Ecophysiology and Stress Physiology of Functional Groups. Springer-Verlag, Berlin. Lee, C. E., J. L. Remfert, and G. W. Gelembiuk. 2003. Evolution of physiological tolerance and performance during freshwater invasions. Integrative and Comparative Biology 43:439-449. Leegood, R. C., and G. E. Edwards. 1996. Carbon metabolism and photorespiration: temperature dependence in relation to other environmental factors. Pp. 191-221 in N. R. Baker, ed. Photosynthesis and the Environment. Kluwer Academic Publishers, Dordrecht, The Netherlands. Lennon, J. J ., J. R. G. Turner, and D. Connell. 1997. A metapopulation model of species boundaries. Oikos 78:486-502. Lesica, P., and F. W. Allendorf. 1995. When are peripheral populations valuable for conservation? Conserv. Biol. 9:753-760. Levin, D. A., and K. Clay. 1984. Dynamics of synthetic Phlox drummondii populations at the species margin. Am. J. Bot. 71 :1040-1050. Lexer, C., M. E. Welch, 0. Raymond, and L. H. Riesebcrg. 2003. The origin of ecological divergence in Helianthus paradoxus (Asteraceae): selection on transgressive characters in a novel hybrid habitat. Evolution 57:1989-2000. Llorens, L., J. Pei’iuelas, C. Beier, B. Emmett, M. Estiarte, and A. Tietema. 2004. Effects of an experimental increase of temperature and drought on the photosynthetic performance of two ericaceous shrub species along a north-south European gradient. Ecosystems 7:613-624. Loik, M. E., and P. S. Nobel. 1993. Freezing tolerance and water relations of Opuntia fi'agilis from Canada and the United States. Ecol 74:1722-1732. Ludwig, F., D. M. Rosenthal, J. A. Johnston, N. Kane, B. L. Gross, C. Lexer, S. A. Dudley, L. H. Rieseberg, and L. A. Donovan. 2004. Selection on leaf ecophysiological traits in a desert hybrid Helianthus species and early- generation hybrids. Evolution 58:2682-2692. 160 Lytle, D. A., and D. M. Merritt. 2004. Hydrologic regimes and riparian forests: a structured population model for cottonwood. Ecology 85:2493-2503. MacArthur, R. H. 1972. Geographical ecology: patterns in the distribution of species. Harper and Row, New York, NY. Machler, F., and J. Nosberger. 1977. Effect of light intensity and temperature on apparent photosynthesis of altitudinal ecotypes of T rifolium repens L. Oecologia 31 :73-78. Mariko, S., H. Koizmni, J .-I. Suzuki, and A. Furukawa. 1993. Altitudinal variations in germination and growth responses of Reynoutriajapom'ca populations on Mt. Fuji to a controlled thermal environment. Ecol. Res. 8:27-34. Maron, J. L., and E. L. Simms. 2001. Rodent-limited establishment of bush lupine: field experiments on the cumulative effect of granivory. J. Ecol. 89:578-588. Marshall, J. K. 1968. Factors limiting the survival of Corynephorus canescens (L.) Beauv. in Great Britain at the northern edge of its distribution. Oikos 19:206- 216. Martin, B., and D. R. Ort. 1985. The recovery of photosynthesis in tomato subsequent to chilling exposure. Photosynth. Res. 6:121-132. Matzner, S., and J. Comstock. 2001. The temperature dependence of shoot hydraulic resistance: implications for stomatal behaviour and hydraulic limitation. Plant Cell Environ. 24: 1299-1307. Maurer, B. A., and M. L. Taper. 2002. Connecting geographical distributions with population processes. Ecol. Lett. 5:223-231. Mayr, E. 1963. Animal species and evolution. Harvard Univ. Press, Cambridge, MA. McClure, M. S., and P. W. Price. 1976. Ecotope characteristics of coexisting Erythroneura leaflroppers (Homoptera; Cicadellidae) on sycamore. Ecology 57:928-940. 161 McKee, J ., and A. J. Richards. 1996. Variation in seed production and germinability in common reed (Phragmites australis) in Britain and France with respect to climate. New Phytol. 133:233-243. McNab, B. K. 1973. Energetics and the distribution of vampires. J. Mammal. 54:131- 143. Medail, F., S. Ziman, M. Boscaiu, J. Riera, M. Lambrou, E. Vela, B. Dutton, and F. Ehrendorfer. 2002. Comparative analysis of biological and ecological differentiation of Anemone palmata L. (Ranunculaceae) in the western Mediterranean (France and Spain): an assessment of rarity and population persistence. Bot. J. Linn. Soc. 140:95-114. Mehlman, D. W. 1997. Change in avian abundance across the geographic range in response to environmental change. Ecol. Appl. 7:614-624. Menges, E. S. 1990. Population viability analysis for an endangered plant. Conserv. Biol. 4:52-62. Menges, E. S., and R. W. Dolan. 1998. Demographic viability of populations of Silene regia in midwestern prairies: relationships with fire management, genetic variation, geographic location, population size, and isolation. J. Ecol. 86:63-78. Miriti, M. N., S. J. Wright, and H. F. Howe. 2001. The effects of neighbors on the demography of a dominant desert shrub (Ambrosia dumosa). Ecol. Monogr. 71 :491-509. Molenaar, F. J ., and A. M. Breeman. 1994. Ecotypic variation in Phyllophora pseudoceranoides (Rhodophyta) ensures winter reproduction throughout its geographic range. J. Phycol. 30:392-402. Maller, A. P. 1995. Patterns of fluctuating asymmetry in sexual ornaments of birds from marginal and central populations. Am. Nat. 145:316-327. Nagy, E. S. 1997. Selection for native characters in hybrids between two locally adapted plant species. Evolution 51:1469-1480. 162 Naidu, S. L., and S. P. Long. 2004. Potential mechanisms of low-temperature tolerance of C4 photosynthesis in Miscanthus x giganteus: an in vivo analysis. Planta 220:145-155. Nantel, P., and D. Gagnon. 1999. Variability in the dynamics of northern peripheral versus southern populations of two clonal plant species, Helianthus divaricatus and Rhus aromatica. J. Ecol. 87:748-760. Nelson, C. J. 1988. Genetic associations between photosynthetic characteristics and yield: review of the evidence. Plant Physiol. Biochem. 26:543-554. Oleksyn, J ., J. Modrzynski, M. G. Tjoelker, R. Zytkowiak, P. B. Reich, and P. Karolewski. 1998. Growth and physiology of Picea abies populations from elevational transects: common garden evidence for altitudinal ecotypes and cold adaptation. Funct. Ecol. 12:573-590. Olmsted, 1., H. Dunevitz, and W. J. Platt. 1993. Effects of freezes on tropical trees in Everglades National Park, Florida, USA. Trop. Ecol. 34:17-34. Orfanidis, S. 1993. Temperature responses and distribution of several Mediterranean macroalgae belonging to different distribution groups. Bot. Mar. 36:359-370. Owens, T. G. 1994. In vivo chlorophyll fluorescence as a probe of photosynthetic physiology. Pp. 195-217 in R. Alscher and A. Wellbum, eds. Plant Responses to the Gaseous Environment. Chapman and Hall, London. Parmesan, C., N. Ryrholm, C. Stefanescu, J. K. Hill, C. D. Thomas, H. Descimon, B. Huntley, L. Kaila, J. Kullberg, T. Tammaru, W. J. Tennent, J. A. Thomas, and M. Warren. 1999. Poleward shifis in geographical ranges of butterfly species associated with regional warming. Nature 399:579-583. Perez-Tris, J ., R. Carbonell, and J. L. Telleria. 2000. Abundance distribution, morphological variation and juvenile condition of robins, Erithacus rubecula (L.), in their Mediterranean range boundary. J. Biogeogr. 27:879-888. Pigott, C. D., and J. P. Huntley. 1981. Factors controlling the distribution of Tilia cordata at the northern limits of its geographical range. 111. Nature and causes of seed sterility. New Phytol. 87:817-839. 163 Pitterman, J ., and R. F. Sage. 2000. Photosynthetic performance at low temperature of Bouteloua gracilis Lag., a high-altitude C4 grass from the Rocky Mountains, USA. Plant Cell Environ. 23:811-823. . 2001. The response of the high altitude C4 grass Muhlenbergia montana (N utt.) A. S. Hitchc. to long- and short-term chilling. J. Exp. Bot. 52:829-838. Prince, S. D. 1976. The effect of climate on grain development in barley at an upland site. New Phytol. 76:377-389. Prince, S. D., and R. N. Carter. 1985. The geographical distribution of prickly lettuce (Lactuca serriola) 111. Its performance in transplant sites beyond its distribution limit in Britain. J. Ecol. 73:49-64. Pulliam, H. R. 1988. Sources, sinks, and population regulation. Am. Nat. 132:652-661. . 2000. On the relationship between niche and distribution. Ecol. Lett. 3:349- 361. Ramsey, J ., H. D. Bradshaw, and D. W. Schemske. 2003. Components of reproductive isolation between the monkeyflowers Mimulus Iewisii and M. cardinalis (Phrymaceae). Evolution 57: 1520-1534. Randall, M. G. M. 1982. The dynamics of an insect population throughout its altitudinal distribution: Coleophora alticolella (Lepidoptera) in northern England. J. Anim. Ecol. 51:993-1016. Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution 43:223-225. Richter, T. A., P. I. Webb, and J. D. Skinner. 1997. Limits to the distribution of the Southern African Ice Rat (0tomys sloggetti): thermal physiology or competitive exclusion? F unct. Ecol. 11:240-246. Root, T. 1988a. Energy constraints on avian distributions and abundances. Ecology 69:330-339. . 1988b. Environmental factors associated with avian distributional boundaries. J. Biogeogr. 15:489-505. 164 Sack, L., P. J. Melcher, M. A. Zwieniecki, and N. M. Holbrook. 2002. The hydraulic conductance of the angiosperm leaf lamina: a comparison of three measurement methods. J. Exp. Bot. 53:2177-2184. Sagarin, R. D., and S. D. Gaines. 2002. The 'abundant centre' distribution: to what extent is it a biogeographical rule? Ecol. Lett. 5:137-147. SASInstitute. 1999. SAS OnlineDoc, Version 8. SAS Institute Inc., Cary, NC. Saxton, A. M. 1998. A macro for converting mean separation output to letter groupings in Proc Mixed. Proceedings of the 23rd SAS Users Group Intl., Nashville, TN Sayed, O. H. 2003. Chlorophyll fluorescence as a tool in cereal crop research. Photosynthetica 41 :321-330. Scheidel, U., S. Rohl, and H. Bruelheide. 2003. Altitudinal gradients of generalist and specialist herbivory on three montane Asteraceae. Acta Oecol. 24:275-283. Schemske, D. W. 1984. Population structure and local selection in Impatiens pallida (Balsaminaceae), a selfing annual. Evolution 38:817-832. Schemske, D. W., and H. D. Bradshaw. 1999. Pollinator preference and the evolution of floral traits in monkeyflowers (Mimulus). Proc. Natl. Acad. Sci. U. S. A. 96:11910-11915. Schreiber, U., W. Bilger, and C. Neubauer. 1994. Chlorophyll fluorescence as a nonintrusive indicator for rapid assessment of in vivo photosynthesis. Pp. 49-70 in ED. Schulze and M. M. Caldwell, eds. Ecophysiology of Photosynthesis. Springer-Verlag, Berlin. Sewell, M. A., and C. M. Young. 1999. Temperature limits to fertilization and early development in the tropical sea urchin Echinometra lucunter. J. Exp. Mar. Biol. Ecol. 236:291-305. Sievert, P. R., and L. B. Keith. 1985. Survival of snowshoe hares at a geographic range boundary. J. Wildl. Manag. 49:854-866. 165 Silberbauer-Gottsberger, I., W. Morawetz, and G. Gottsberger. 1977. Frost damage of cerrado plants in Botucatu, Brazil, as related to the geographical distribution of the species. Biotropica 9:253-261. Sjogren, P. 1991. Genetic variation in relation to demography of peripheral pool frog populations (Rana lessonae). Evol. Ecol. 5:248-271. Slatyer, R. O. 1977. Altitudinal variation in the photosynthetic characteristics of snow gum, Eucalyptus pauciflora Sieb. ex Spreng. III. Temperature response of material grown in contrasting thermal environments. Aust. J. Plant Physiol. 42301-3 12. Stanton, M. L., and C. Galen. 1997. Life on the edge: adaptation versus environmentally mediated gene flow in the Snow Buttercup, Ranunculus adoneus. Am. Nat. 150:143-178. Stokes, K. E., J. M. Bullock, and A. R. Watkinson. 2004. Population dynamics across a parapatric range boundary: Ulex gallii and Ulex minor. J. Ecol. 92: 142-155. Svensson, B. W. 1992. Changes in occupancy, niche breadth and abundance of 3 Gyrinus species as their respective range limits are approached. Oikos 63: 147- 156. Taniguchi, Y., and S. Nakano. 2000. Condition-specific competition: implications for the altitudinal distribution of stream fishes. Ecology 81:2027-2039. Telleria, J. L., and T. Santos. 1993. Distributional patterns of insectivorous passerines in the Iberian forests: Does abundance decrease near the border? J. Biogeogr. 20:23 5-240. Terborgh, J ., and J. S. Weske. 1975. The role of competition in the distribution of Andean birds. Ecology 56:562-576. Thomas, C. D., E. J. Bodsworth, R. J. Wilson, A. D. Simmons, Z. G. Davies, M. Musche, and L. Conradt. 2001. Ecological and evolutionary processes at expanding range margins. Nature 411:577-581. Turesson, G. 1922. The genotypical response of the plant species to the habitat. Hereditas 3:211-350. 166 Tumbull, L. A., M. J. Crawley, and M. Rees. 2000. Are plant populations seed-limited? A review of seed sowing experiments. Oikos 88 Tyree, M. T., S. Patino, J. Bennink, and J. Alexander. 1995. Dynamic measurements of root hydraulic conductance using a high-pressure flowmeter in the laboratory and field. J. Exp. Bot. 46:83-94. Van Rossum, F., X. Vekemans, E. Gratia, and P. Meerts. 2003. A comparative study of allozyme variation of peripheral and central populations of Silene nutans L. (Caryophyllaceae) from Western Europe: implications for conservation. Plant Syst. Evol. 242:49-61. Verhoeven, K. J. F., T. K. Vanhala, A. Biere, E. Nevo, and J. M. M. Van Damme. 2004. The genetic basis of adaptive population differentiation: a quantitative trait locus analysis of fitness traits in two wild barley populations from contrasting habitats. Evolution 58:270-283. Vickery, R. K., Jr. 1967. Experimental hybridizations in the genus Mimulus. VI. Section Erythranthe. Proceedings Utah Academy of Sciences 44:321-333. . 1978. Case studies in the evolution of species complexes in Mimulus. Evol. Biol. 11:404-506. Volis, S., S. Mendlinger, and D. Ward. 2004. Demography and role of the seed bank in Mediterranean and desert populations of wild barley. Basic and Applied Ecology 5:53-64. Vucetich, J. A., and T. A. Waite. 2003. Spatial patterns of demography and genetic processes across the species' range: Null hypotheses for landscape conservation genetics. Conservation Genetics 4:639-645. Waser, N. M., R. K. J. Vickery, and M. V. Price. 1982. Patterns of seed dispersal and population differentiation in Mimulus guttatus. Evolution 36 Williams, C. K., A. R. Ives, and R. D. Applegate. 2003. Population dynamics across geographical ranges: Time-series analyses of three small game species. Ecology 84:2654-2667. 167 Woodward, F. I. 1975. The climatic control of the altitudinal distribution of Sedum rosea (L.) Scop. and S. telephium L. 11. The analysis of plant grth in controlled environments. New Phytol. 74:335-348. . 1979. The differential temperature responses of the grth of certain plant species from different altitudes. 1. Growth analysis of Phleum alpinum L., P. bertolonii D. C., Sesleria albicans Kit. and Dactylis glomerata L. New Phytol. 82:385-395. . 1990. The impact of low temperatures in controlling the geographical distribution of plants. Philos. Trans. R. Soc. Lond. B Biol. Sci. 326:585-592. Woodward, F. I., and C. D. Pigott. 1975. The climatic control of the altitudinal distribution of Sedum rosea (L.) Scop. and S. telephium L. 1. Field observations. New Phytol. 74:323-334. Xiong, F. S., C. T. Ruhland, and T. A. Day. 1999. Photosynthetic temperature response of the Antarctic vascular plants Colobanthus quitensis and Deschampsia antarctica. Physiol. Plant 106:276-286. Zacherl, D., S. D. Gaines, and S. I. Lonhart. 2003. The limits to biogeographical distributions: insights from the northward range extension of the marine snail, Kelletia kelletii (Forbes, 1852). J. Biogeogr. 30:913-924. 168 lllilillflijlfijljljjillljjljijflu