University of Florida Thesis Or Dissertation

SEED MASS AND DETERIORATION: IMPLICATIONS FOR IN SITU SURVIVAL AND EX SITU LONGEVITY

NICHOLAS GERARD GENNA

A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2019

To Ricky and Franklin and Arnold and Albert

ACKNOWLEDGMENTS

I firstly thank my family for their continual and unwavering support. I especially thank my advisor, Héctor Pérez, for his persistent encouragement and motivation. I would not have achieved success as an undergraduate or graduate researcher without your mentorship. I also thank my dissertation committee including Jamie Gillooly, Alfred

Huo, Michael Kane, and Christina Walters for challenging my proposed work and for providing thoughtful insight. I want to recognize the years of support and constructive criticism of lab members Amber Gardener and Tia Tyler. I also want to acknowledge the technical and resource support provided by Terry Zinn of Wildflowers of Florida Inc. and

Karen Kelly and Rudy Alvarado of the UF ICBR EM core. Finally, I want to recognize the support of Alec Chin-Quee: thank you for the years of technical support and for giving me the opportunity to lead you in the beginning of your research career.

TABLE OF CONTENTS

page

ACKNOWLEDGMENTS ...... 4

LIST OF TABLES ...... 8

LIST OF FIGURES ...... 10

ABSTRACT ...... 12

CHAPTER

1 SEED MASS VARIATION AND ITS ECOLOGICAL EFFECTS ...... 14

Problem Statement ...... 14 Seed Deterioration ...... 17 Evidence of Mass-Dependent Intraspecific Seed Deterioration ...... 22 Conclusion ...... 23 Hypotheses ...... 25 Hypothesis 1: Inherent Physical and Physiological Differences Will Correlate with Seed Mass and Manifest from Developmental Location...... 25 Hypothesis 2: Higher Mass Seeds Will Deteriorate at a Slower Rate During In Situ Burial Compared to Lower Mass Seeds...... 27 Hypothesis 3: Seed Mass Will Promote Differential Deterioration In Ex Situ Storage...... 28 Experimentally Demonstrating Differential Viability Loss of Variable Mass Seeds In Situ and Ex Situ ...... 28

2 CHARACTERIZING GERMINATION ECOLOGY, DEVELOPMENT, DISTRIBUTION, AND INTRACELLULAR ANATOMY IN RUDBECKIA MOLLIS (ASTERACEAE) SEEDS OF DIFFERENT MASS ...... 34

Introduction ...... 34 Materials and Methods...... 35 Plant Material and Establishing Mass Classes ...... 35 Germination of Fresh Seeds ...... 37 Greenhouse Study ...... 37 Plant establishment ...... 37 Developmental study ...... 38 Seed mass distribution study ...... 39 Comparative Seed Anatomy ...... 40 Seed Density ...... 41 Statistical Analyses ...... 41 Germination testing ...... 41 Seed mass distribution ...... 42 Developmental study, comparative seed anatomy, true density ...... 43

Results ...... 43 Germination Testing When Fresh ...... 43 Developmental Study ...... 44 Seed Mass Distribution ...... 45 Comparative Seed Anatomy and Density ...... 48 Discussion ...... 49 Germination Testing When Fresh ...... 49 Developmental Study and Distribution of Seed Mass ...... 51 Comparative Seed Anatomy and Density ...... 53 Conclusion ...... 54

3 DOES INTRA-POPULATION SEED SURVIVAL DURING BURIAL FOLLOW INTERSPECIFIC RELATIONSHIPS BETWEEN SEED MASS AND COAT THICKNESS? ...... 71

Introduction ...... 71 Materials and Methods...... 74 Plant Material and Establishing Mass Classes ...... 74 Seed Burial ...... 74 Germination Testing Following Burial ...... 75 Allocation to Physical Defenses ...... 75 Statistical Analyses ...... 76 Results ...... 77 Air and Soil Climate ...... 77 Seed Count Data ...... 77 Germination Following Burial ...... 78 Allocation to Physical Defenses ...... 79 Discussion ...... 80 Conclusion ...... 84

4 NO EVIDENCE OF MASS DEPENDENT SEED LONGEVITY FOLLOWING STORAGE OR REPEATED RELATIVE LONGEVITY ASSESSMENTS ...... 95

Introduction ...... 95 Materials and Methods...... 100 Plant Material and Establishing Mass Classes ...... 100 Water Sorption Isotherms ...... 100 Seed Storage Treatments ...... 101 Saturated Salt Accelerated Aging (SSAA) ...... 102 Statistical Analyses ...... 103 Results ...... 103 Water Sorption Isotherms ...... 103 Seed Storage Treatments ...... 103 Saturated Salt Accelerated Aging ...... 104 Discussion ...... 105 Conclusion ...... 110

5 CONCLUSIONS ...... 119

LIST OF REFERENCES ...... 127

BIOGRAPHICAL SKETCH ...... 139

LIST OF TABLES

Table page

1-1 Overall Cox regression model for the germination response of Rudbeckia mollis seeds of different mass to increasing saturated salt accelerated aging duration...... 30

1-2 Linear contrasts for germination of Rudbeckia mollis mass classes to increasing duration of saturated salt accelerated aging ...... 31

2-1 Cox regression models for the germination response of fresh Rudbeckia mollis seeds of different mass exposed to simulated seasonal alternating and constant temperatures...... 55

2-2 Orthogonal contrast comparisons for germination of Rudbeckia mollis mass classes to seasonal and constant temperatures controlling for seed mass ...... 56

2-3 Orthogonal contrast comparisons for germination of Rudbeckia mollis mass classes within seasonal temperatures ...... 57

2-4 Rudbeckia mollis plant characteristics ...... 58

2-5 Four level hierarchical ANOVA assessing seed mass variation within Rudbeckia mollis plants ...... 59

2-6 Spearman correlation coefficients and probability statistics of mean inflorescence seed mass and measured plant characteristics ...... 60

2-7 Spearman correlation coefficients and probability statistics of individual seed masses and measured plant characteristics ...... 61

2-8 Anatomical measurements from different locations within Rudbeckia mollis seeds of different mass ...... 62

3-1 Nonzero and general association statistics for filled, empty, and total seed counts of Rudbeckia mollis seeds buried for 24 months...... 85

3-2 Spearman correlation statistics for filled, empty, and total seed counts correlated with Rudbeckia mollis mass classes buried for 24 months...... 86

3-3 Cox regression model for the germination response of Rudbeckia mollis seeds of different mass following 24 months of burial ...... 87

3-4 Selected orthogonal contrast comparisons of Rudbeckia mollis mass classes within burial months ...... 88

3-5 Pearson correlation statistics for Rudbeckia mollis final germination percentage and germination rate correlated with soil temperature following burial for 24 months...... 89

4-1 Final germination percentages for Rudbeckia mollis mass classes following equilibration to various relative humidity environments ...... 111

4-2 Deterioration parameters for Rudbeckia mollis mass classes when fresh and on month 0 of the storage experiment determined from the Ellis and Roberts viability equation...... 112

4-3 Deterioration parameters for Rudbeckia mollis mass classes determined from the Ellis and Roberts viability equation following 24 months of storage in a shed and in a room...... 113

LIST OF FIGURES

Figure page

1-1 Kaplan-Meier survival functions for Rudbeckia mollis mass classes following saturated salt accelerated aging ...... 32

1-2 Germination parameters for Rudbeckia mollis mass-classes following 0-30 days of saturated salt accelerated aging (SSAA) ...... 33

2-1 Example of floret sampling procedure for the seed development and seed mass distribution study ...... 64

2-2 Germination response of fresh Rudbeckia mollis mass classes exposed to simulated seasonal alternating temperatures ...... 65

2-3 Germination response of fresh Rudbeckia mollis mass classes to constant temperatures ...... 66

2-4 Rudbeckia mollis inflorescences during development ...... 67

2-5 Rudbeckia mollis seeds sampled following anthesis ...... 68

2-6 Rendering of Rudbeckia mollis ...... 69

2-7 Light and transmission electron microscope images of Rudbeckia mollis seeds ...... 70

3-1 Air and soil conditions in the burial plot measured from January 1 to December 31, 2018 ...... 90

3-2 Counts of Rudbeckia mollis seeds from the light, intermediate, and heavy mass classes following 24 months of burial ...... 91

3-3 Final germination percent data for Rudbeckia mollis mass classes following 24 months of burial ...... 92

3-4 Germination summary statistics for Rudbeckia mollis mass classes following 24 months of burial ...... 93

3-5 Relationship between pericarp thickness and seed mass of Rudbeckia mollis seeds separated into a light, intermediate, and heavy mass class ...... 94

4-1 Water sorption isotherms for Rudbeckia mollis mass classes following equilibration to various relative humidity environments ...... 114

4-2 Climate inside the non-climate controlled outdoor shed from January 1 to December 31, 2018 ...... 115

4-3 Final germination percentage for Rudbeckia mollis mass classes following storage for 24 months ...... 116

4-4 Final germination percentages of Rudbeckia mollis mass classes subjected to saturated salt accelerated aging stress for 45 days at 41.0oC ...... 117

4-5 Final germination percentages of Rudbeckia mollis mass classes stored in a shed and room and subjected to saturated salt accelerated aging at 41.0oC for 45 days at different time points ...... 118

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

SEED MASS AND DETERIORATION: IMPLICATIONS FOR IN SITU SURVIVAL AND EX SITU LONGEVITY

Nicholas Gerard Genna

August 2019

Chair: Héctor E. Pérez Major: Horticultural Sciences

Previous research demonstrated potential mass-based longevity differences in

Rudbeckia mollis (Asteraceae) following high temperature aging. However, few empirical studies have explored intra-population seed mass variation in the context of ex situ storage and in situ burial. This research attempted to characterize the anatomy and physiology of different mass seeds, pinpoint their developmental location within mother plants, and understand how these differences may modulate deterioration under different environmental conditions. Seeds were harvested during August 2016 and separated into a light (383 ± 7 µg), intermediate (409 ± 8 µg), and heavy (447 ± 10 µg) mass class. Germination was independent of mass under constant 15-35oC, simulated spring (29/19oC), and summer (33/24oC) temperatures when fresh. Simulated fall

(27/15oC) and winter (22/11oC) temperatures revealed mass-dependent conditional dormancy in heavy class seeds while light class seeds were non-dormant. A developmental study demonstrated that seeds reached mass maturity at or before four weeks following anthesis and are dehiscent by week eight. A seed mass distribution study showed seed mass was greatest in higher order primary branches, lower order secondary branches, and in proximal positions within inflorescences. Seed mass

variation was greatest within plants with 40.3% of that variation occurring within inflorescences. Germination fluctuated from 28-98% during a two year burial study indicative of dormancy cycling and consistent with non-deep physiological dormancy.

Higher mass seeds with thicker seed coats were more likely to be recovered filled during burial due to greater tolerance to pathogen pressure. Finally, seed viability remained ≥ 81% across classes for two years when stored in a room (23oC, 42.5% relative humidity (RH)), at 4oC (15% RH), or -18oC (15% RH) while viability fell to 11% in a non-climate controlled shed. Repeated high temperature and high humidity aging assays tentatively predicted mass-dependent longevity in the shed while highlighting declining vigor in room stored seeds. Overall, seed mass does not drive longevity during ex situ storage but may correlate with in situ survival. I argue that this research with R. mollis offers no empirical support for the intra-population mass-dependent deterioration hypothesis purported by high temperature aging.

CHAPTER 1 SEED MASS VARIATION AND ITS ECOLOGICAL EFFECTS

Problem Statement

In this study, I intend to show that intraspecific seed mass variation is a primary factor determining in situ survival and ex situ longevity. I will also address the scientific knowledge gap concerning the ecological and evolutionary significance of intraspecific seed mass variation in wild plants by promoting understanding of: 1) intraspecific seed mass variation and germination ecology, 2) how intraspecific seed mass variation may drive viability loss during burial in soil, and 3) how seed mass variation can be used to enhance current ex situ storage protocols.

To date, interspecific, rather than intraspecific seed mass studies dominate the literature. One goal of interspecific seed mass research is to understand why species, differing in lineage and functional traits, evolved vastly different mean seed sizes in a common environment (Moles and Westoby, 2004). Researchers are also attempting to understand why there is a species-specific genetic component of longevity in ex situ seed banks (Walters et al., 2005). Unfortunately, interspecific seed mass research neglects the marked intraspecific mass variation that occurs in wild plant accessions.

Unlike domesticated counterparts that experienced intense breeding pressure for uniformity, wild plants possess greater intraspecific seed mass variation due to a need to respond to a multitude of stochastic events. Seed mass variation potentially increases plant fitness through bet hedging, where a subset of seeds are capable of high germination and establishment in good years, while another subset of seeds decrease local extinction risk via increased persistence and germination avoidance in bad years (Thompson, 2000). While beneficial in the wild, intraspecific seed mass

variation reduces storage behavior uniformity and complicates the efficacy of conservation efforts for imperiled plant species (Hay and Probert, 2013; Walters, 2015).

Understanding the factors that drive non-uniform storage behavior in wild seeds will improve their storability and reduce costs for native plant growers and large seedbanks.

This can be accomplished by linking seed phenotypes to quantifiable functional traits, including seed mass. For this reason, the current research will highlight intraspecific seed mass variation, an often overlooked functional trait in plants in regards to deterioration, and determine the extent to which this phenomenon contributes to differential in situ and ex situ viability loss.

Intraspecific seed mass variation is a product of constraints on seed development. Original theoretical models predicted that equal offspring investment maximizes both parental and offspring fitness (McGinley et al., 1987; Smith and

Fretwell, 1974). Therefore, stabilizing selection for seed size should manifest in one optimal seed size for each species in a given environment. However, intraspecific variation in seed mass is widely observed in both gymnosperm (Righter, 1945) and angiosperm (Obeso, 1993; Vaughton and Ramsey, 1997; Winn, 1991) lineages.

McGinley et al. (1987) argues that although some optimal seed mass is predicted for each individual, maternal and environmental effects prevent uniform seed production.

For example, seed mass varies within a single inflorescence (Torices and Mendez,

2010), among inflorescences on a flowering branch (Hendrix, 1984), or between plants within a population (Baloch et al., 2001). Within an inflorescence, seed mass decreases acropetally due to either resource limiting effects (Stephenson, 1981), architecture effects (Diggle, 1995), or a combination of both (Torices and Mendez, 2010). Most

intraspecific seed mass variation has been attributed to within inflorescence origin

(Vaughton and Ramsey, 1997; Winn, 1991). Therefore, it is conceivable that the lack of uniform offspring provisioning within inflorescences, among inflorescences, and between plants within a population, may produce seeds of variable fitness identifiable as mass-specific phenotypes.

Interestingly, mass-induced fitness differences are widely demonstrated in the literature. For instance, seeds of greater mass tend to germinate to higher percentages, germinate at faster rates, and display higher recruitment compared to lower mass seeds within species (Afonso et al., 2014; Dai et al., 2018; Hendrix, 1984; Hendrix and Trapp,

1992; Khan, 2004; Khan and Ungar, 1984; Kolodziejek, 2017; Larsen and Andreasen,

2004; Lehtila and Ehrlen, 2005; Moravcova et al., 2005; Nelson and Johnson, 1983;

Schaal, 1980; Stanton, 1984; Torices et al., 2013; Tripathi and Khan, 1990; van Molken et al., 2005; Wang et al., 2010; Weis, 1982; Wulff, 1973; Zimmerman and Weis, 1983).

This research indicates that a positive correlation between intraspecific seed mass variation and germination ability may exist in most species (Baskin and Baskin, 2014).

Yet, for other species, germination percent, rate, or recruitment was independent

(Guillemin and Chauvel, 2011; Hernandez and Orioli, 1985; Parker et al., 2006; Torices and Mendez, 2010), decreased (Banovetz and Scheiner, 1994), or increased and decreased with mass (Hou and Romo, 1998; Marshall, 1986; Milberg et al., 1996). The conservation of intraspecific seed mass variation in nature, and the lack of consistent germination response as a function of seed mass across species, suggests a tradeoff may exist between seed mass and some other unrecognized trait.

Here, I argue that seed longevity, or the maintenance of viability over time, may be one unrecognized seed trait in wild plants that is directly related to intraspecific seed mass variation in wild plants. Seed viability is defined as the ability of a seed to complete the germination process that ends with radicle emergence (Probert and

Linington, 2006). Inherent seed longevity can vary temporally and individually, but longevity is primarily conserved at the species level. For example, some species possess greater inherent in situ and ex situ longevity than other species (Long et al.,

2008; Walters et al., 2005). Currently, more research exists concerning the effect of intraspecific seed mass variation on the germination process than the effect of seed mass on viability loss. For example, only Banovetz and Scheiner (1994), Espinosa-

Garcia et al. (2003), and Tricault et al. (2017) documented the longevity of mass- separated seeds buried in the soil. I found no papers that directly assessed the effect of intraspecific seed mass variation on longevity during long term storage. Intraspecific seed mass variation is ubiquitous in plants. Therefore, research is needed to understand: 1) how seeds of variable mass interact with their post-dispersal environment under various conditions of abiotic stress and 2) how intraspecific seed mass variation within wild plant accessions may modulate storage behavior in ex situ seed banks.

Seed Deterioration

Seeds lose viability through a natural and fundamentally unresolved process called aging. Aging is quantified in seeds as viability and vigor loss in standard germination tests (Matthews et al., 2012). For example, mean time to germination will increase and germination percentage will decrease as seeds age. Under cold and dry storage conditions, seed deterioration proceeds in a negative sigmoid fashion with three

distinct phases. Seeds are asymptomatic of deterioration during phase one. Seed lots at different points in phase one are difficult to distinguish because germination is high and normal. Seed longevity is defined as the duration of phase one and is quantified by seed lot viability ≥ 80% in ex situ storage (Walters, 2014). However, at some threshold time, seeds enter phase two of the deterioration curve and viability loss occurs rapidly.

Seeds in phase two of the deterioration process become undesirable for long term storage as viability loss is accelerated and field germination becomes improbable

(Walters, 2015). Therefore, the goal of long-term ex situ seed storage is to maintain seeds in phase one of the deterioration curve and to regenerate accessions if phase two becomes imminent. Finally, phase three begins when the precipitous decrease in viability slows and all viability is eventually lost.

Factors that contribute to aging include seed moisture, temperature, and time

(Ellis and Roberts, 1980; Walters, 1998). Maturation drying represents the final phase of seed development. During maturation drying, desiccation tolerant seeds lose intracellular water, but some cellular water remains tightly associated with intracellular macromolecules forming amorphous solids called rubbers and glasses. Rubbers and glasses are formed in dry seeds and are distinguished by their degree of molecular spacing and mobility; with glasses forming at relatively lower water contents and restricting molecular movement the greatest. Although glass formation occurs naturally in desiccation tolerant seeds at temperatures above freezing, ex situ storage reinforces natural glass formation by exposure to -18oC. Low water content and freezing conditions work in tandem to restrict molecular movement and make deteriorative processes thermodynamically unfavorable. Together, temperature and water content

are the two most important factors that regulate aging in cold and dry storage (Walters,

2005). However, cold and dry conditions cannot stop intracellular molecular movement and all seeds eventually lose viability as glassy structures collapse, recrystallize, and relax promoting molecular contact and deteriorative processes to occur (Walters, 2014).

Ultimately, these reactions manifest in viability loss during standard germination testing.

In contrast to artificially dry conditions experienced by seeds ex situ, seed water content in situ will equilibrate with soil water potential or ambient air relative humidity

(RH) depending on a seed’s location. Seeds may potentially counteract the deleterious effects of accumulated aging by-products through biochemical and structural repair processes via periodic hydration (Long et al., 2011; Wojtyla et al., 2016). Therefore, physiological aging may only play a minor role while pathogen or predator pressure may play major roles in promoting viability loss in situ in the short term (Long et al., 2015).

This is not to say, however, that temperature, moisture status, and time do not play any or only minor roles during the deterioration process in situ in the long term, as seeds can still age and die while not suffering any pathogen pressure in soil. Over time, seed coat fractures may form permitting fungal and bacterial pathogens to enter and consume nutritive or embryo tissue. Ultimate seed death is likely once physical seed defenses are lost. Furthermore, the pattern of viability loss in situ is not resolved to the same extent as viability loss ex situ, and it is unclear whether seeds lose viability in a negative sigmoid fashion. On the other hand, persistence in soil occurs in a negative exponential fashion (Murdoch and Ellis, 2000). However, persistence in this case includes germination from the soil seed bank along with viability loss and death due to

predation. Therefore, the negative exponential conclusion does not apply to longevity in soil.

Examples of long-term empirical seed storage experiments are not abundant due to the incredible investment of time and resources required. High temperature aging assays are used to circumvent this issue and are referred to as “accelerated aging,”

“saturated salt accelerated aging,” or “controlled deterioration” in the literature depending on conditions imposed. High temperature aging was first used to determine the relative storability of similar seed lots of unknown quality (Delouche and Baskin,

1973). Today, high temperature aging is utilized as a standardized vigor test for crops or as a comparative assay of longevity (AOSA, 2002; Jianhua and McDonald, 1996;

Newton et al., 2009). Seed viability loss is accelerated by exposing seeds to a high temperature and high RH environment for an extended period. The elevated RH causes a direct increase in seed water content and the high temperature accelerates molecular mobility and reactive oxygen species generation. Seed water content, however, remains below the threshold needed for metabolism to occur, thereby keeping seeds in a state of deterioration without the ability to respond with significant repair.

Saturated salt accelerated aging (SSAA) is a modified version that utilizes various saturated salt solutions in place of pure water. SSAA is most appropriate for smaller seeded species that may experience a rapid water content increase and corresponding rapid deterioration that would mask differences between seed lots of interest. Also, the lower seed water content reduces or eliminates fungal growth

(Jianhua and McDonald, 1996). There is current debate concerning the applicability of high temperature aging research to the study of deterioration in ex situ storage, since

one would have to assume that the type of physiological reactions occurring in a hot and moist environment would also occur during cold and dry storage (Walters et al.,

2005). This assumption is likely false (Ballesteros and Walters, 2011) and has spurred the development of new aging techniques such as subjecting seeds to compressed air or compressed pure oxygen treatments (Groot et al., 2012; Nagel et al., 2016).

Exposing seeds to increased oxygen partial pressure mimics accelerated deterioration kinetics under high temperature and RH conditions. However, elevated partial pressure of oxygen aging has the presumed advantage of reducing seed viability while maintaining an intracellular glassy state.

In contrast to high temperature aging, longevity in situ has been historically simulated by burying seeds under field conditions in permeable bags. Over time, bags are exhumed and seeds are tested for viability with a standard germination assay. Since germination is deterred by burial depth, seed burial studies involving bags are specifically testing the longevity of seeds within the soil seed bank at the specific depth of burial. Researchers conducting these experiments are solely interested in identifying, under natural conditions, the length of time seeds will remain viable within soil.

However, a distinction should be made between seed burial within bags and phenology studies. In the latter, researchers broadcast freshly harvested seeds on soil and monitor emergence over time to understand when germination is occurring in nature.

Researchers are not interested in longevity in that case, or the period that a seed will remain viable in soil under natural conditions. Rather, researchers are interested in persistence, or the tendency for seeds to avoid germination due to the lack of a specific germination cue or dormancy (Long et al., 2015). Therefore, seed burial studies within

bags ensure seeds will remain persistent within the soil by artificial burial to a specific depth, but results cannot be used as a correlate for persistence since germination is deterred by burial depth.

Evidence of Mass-Dependent Intraspecific Seed Deterioration

My primary supposition that seed mass and deterioration are linked is founded on intraspecific seed mass research with Rudbeckia mollis (Genna and Pérez, 2016).

Rudbeckia mollis seeds were collected at maturity by a native plant grower in North

Central Florida during September 2013. Seeds were fractionated with an air density separator and fractions were combined into three mass-based classes called light, intermediate, and heavy. Mean seed mass in each class was 393 ± 35, 423 ± 29, and

474 ± 38 μg (mean ± SE), respectively. Each mass class was first subjected to standard germination tests under a range of alternating (22/11, 27/15, 29/19 and 34/24oC) and constant (25, 27.5, 30.0, 32.5, and 35oC) temperature regimes. These tests demonstrated that inherent germination ability was independent of mass. Next, mass classes were subjected to saturated salt accelerated aging (SSAA) to determine if a mass-dependent germination response would become evident following stress. SSAA was carried out by suspending seeds over a saturated NaCl solution (75% RH, -39

MPa) in a 41.0oC incubator for 0-30 days. Viability was assessed after all SSAA durations and for each mass class via a standard germination test at 25.0oC for 28 days.

From this research, Genna and Pérez (2016) found temporal patterns of germination were similar between mass classes following 0 and 5 d of SSAA.

Germination began to slow after 10 d and the probability of not germinating began increasing to an average of 0.27 across classes by 15 d. Temporal patterns of

germination became distinct by 20 d with the probability of not germinating in the light class increasing slightly to 0.32 while intermediate and heavy classes increased dramatically to above 0.55. Viability was significantly reduced following 25 d SSAA and only one seed germinated in the light class following 30 d SSAA (Figure 1-1). The total number of germinated seeds was 88-93 in all classes between 0-10 d of SSAA. Total germination decreased considerably following 15-30 d SSAA in all classes. However, light class seed germination count was 1.7 times higher than the average of intermediate or heavy class seeds following 20 d SSAA (Figure 1-2A). Differences in germination rate were also evident following SSAA. Light class seeds germinated relatively faster than intermediate or heavy seeds following 5 and 10 d SSAA (Figure 1-

2B). We rejected the hypothesis that survivor functions were similar across SSAA

2 durations and mass classes (Log-Rank  20 = 1773.65; P <0.0001).

Cox regression indicated that both mass class and SSAA duration significantly influenced germination response (Table 1-1). Linear contrasts showed that light class seeds germinated significantly better than intermediate or heavy class seeds after 10 and 20 days of SSAA (Table 1-2). Visually, light class germination was only slightly better after 10 days of SSAA in comparison to the dramatic difference in germination response after 20 days of SSAA (Figure 1-1).

Conclusion

For the first time, the experiment described above demonstrated that mass separated R. mollis seeds tolerate aging stress differently. However, two questions arose following the completion of this research that require further investigation. The first question concerns the degree to which fresh seeds will exhibit similar mass-

dependent germination responses following SSAA. Seeds used in the study above were stored at 4oC for two years before SSAA was carried out. Therefore, it is debatable if the aging stress imposed was the principal cause of deterioration, or if high temperature aging stress simply revealed differences among seeds that were already present. The likelihood exists that SSAA revealed differences between mass classes that developed in storage, since this is the prescribed use of accelerated aging as a relative vigor test

(AOSA, 2002).

The second question that is evident upon further consideration is if similar results would manifest in ex situ storage or during in situ burial. In situ, soil water content may vary considerably from completely dry to completely wet depending upon precipitation, evaporation, and matric forces of soil particles. During this natural fluctuation, seeds may be exposed to a 75% RH environment similar to conditions that are imposed by

NaCl during SSAA. However, exposure to a 75% RH environment may be short and fleeting as water enters and leaves the soil. In contrast, seed water content is equilibrated in a 15% RH environment prior to ex situ storage, which is well below the

75% RH provided by NaCl during SSAA (FAO, 2014). Therefore, it is plausible that aging results obtained under a 75% RH environment imposed by SSAA may not mirror either an in situ or ex situ environment. Nevertheless, results presented above provide motivation to test hypotheses related to intraspecific seed mass variation and deterioration in an ex situ and in situ environment through direct seed storage and burial experiments.

From the perspective of wild plants, intraspecific seed mass variation may be driven by a need to modulate germination ability and seed longevity to increase fitness

and survivorship during good and bad seed producing years. In wild plants, seeds are produced on a continuum of seed mass: some seeds are larger and heavier while others are smaller and lighter. Generally, higher mass seeds have a germination advantage and may serve as primary establishing individuals during the season that emergence is typical. In contrast, lower mass seeds may have a disposition for persistence, effectively waiting on or in the soil seed bank for permissible conditions to emerge. Therefore, lower mass seeds may serve as primary placeholders within the soil seed bank with the principal purpose of emerging years later. This “general ecological model for intraspecific seed mass variation” accounts for many of the observed trends in intraspecific seed mass germination research by incorporating potential longevity differences from the SSAA experiment described previously. Under this model, intraspecific seed mass variation may dictate seed roles in nature.

On the other hand, and in the realm of ex situ seed storage, greater longevity of one mass fraction compared to others from the same seed lot could undermine long term seed storage goals. Understanding how intraspecific seed mass variation potentially modulates aging-related seed deterioration can help explain the marked differences in storability within wild plant accessions discussed by Walters (2015).

Furthermore, genetic variation can be optimized within wild accessions by carrying out regeneration work while whole seed lot viability is higher and before a subset of seeds show early signs of deterioration.

Hypotheses

Hypothesis 1: Inherent Physical and Physiological Differences Will Correlate with Seed Mass and Manifest from Developmental Location.

Genna and Pérez (2016) showed a mass-based germination response in R. mollis seeds following SSAA. For this to occur, inherent physical and physiological differences must exist in different mass seeds to drive differential deterioration. For example, seeds of different mass may differ in initial germination ability due to mass- dependent dormancy or quality differences. However, current evidence suggests that initial seed germination does not correlate with long term viability (Walters et al., 2005).

Also, relative proportions of proteins, lipids, or starch could differ with mass.

Macronutrients, particularly lipids, directly affect water relations through an inverse relationship between lipid content and seed water content. If lipid content is mass dependent, seeds of different mass will equilibrate to dissimilar water contents and lose viability at different rates. Similarly, cellular reserve accumulation could differ with mass.

Cellular reserves deposited during seed development confer desiccation tolerance by preventing plasma membranes from becoming mechanically disturbed (Walters et al.,

2002; Walters and Koster, 2007). Current literature suggests that a minimum threshold of 35% intracellular dry matter is needed for seeds to tolerate extreme drying (Pérez et al., 2012). Assuming R. mollis seeds exhibit orthodox storage behavior, and that intracellular dry matter exceeds 35% in each mass class, small differences between seeds of different mass may manifest into quantifiable mass-dependent viability loss during post-storage germination assays. Finally, all factors described above could be driven by developmental location on mother plants. Most wild plants produce seeds of different mass due to resource effects. Seeds developing closer to vascular connections are stronger resrouce sinks and will have greater mass than seeds developing distal from primary vasculature. Ultimately, differences in macronutrient composition, initial

germination ability, and desiccation tolerance can be linked back to developmental location with a comprehensive developmental study documenting seed development in situ and in planta seed mass distribution.

Hypothesis 2: Higher Mass Seeds Will Deteriorate at a Slower Rate During In Situ Burial Compared to Lower Mass Seeds.

A seed’s ability to maintain viability during long term seed burial in situ will depend on physiological aging and the rate of repair and pathogen or predator pressure. In soil, seeds will maintain viability for the duration that physical defenses, such as testa or pericarp tissue, remain intact providing a physical barrier to pathogen intrusion. In the short term, physiological aging may play a minor role in viability loss as transient hydration may permit the repair of aging damage (Long et al., 2015).

Therefore, pathogen pressure may primarily determine the rate of viability loss in soil. I propose the supposition that higher mass seeds will resist pathogen attack to a greater extent than lower mass seeds due to thicker seed coats. This relationship was demonstrated interspecifically by Gardarin et al. (2010) and intraspecifically (inter- accession) by Schutte et al. (2014). Both authors found mortality in soil decreased with increasing seed coat thickness. Genna (2015) found a direct increasing relationship between seed mass and pericarp thickness in R. mollis. Greater pericarp thickness in higher mass R. mollis seeds may offer better resistance to pathogen pressure while buried in soil. Alternatively, lower mass seeds may experience more rapid viability loss due to thinner seed coats. A similar conclusion was also found by Banovetz and

Scheiner (1994) who showed that higher mass Coreopsis lanceolata (Asteraceae) seeds retained significantly higher viability following two years of burial compared to

lower mass seeds. Unfortunately, Banovetz and Scheiner (1994) did not measure pericarp thickness and no such study is published on C. lanceolata.

Hypothesis 3: Seed Mass Will Promote Differential Deterioration In Ex Situ Storage.

Ex situ seed longevity is directly related to a seed’s inherent ability to resist physiological aging stress while stored in a controlled environment. To date, I have not found any research that has examined the relationship between intraspecific seed mass variation and deterioration during ex situ storage. However, studies have overwhelmingly shown that there is no correlation between interspecific seed mass variation and longevity in storage. For example, Walters et al. (2005) examined storage longevity for 276 species stored at the National Center for Genetic Resources

Preservation for up to 80 years. This research concluded that individual species may exhibit characteristic storage longevities but that seed mass at the species level does not correlate with longevity. However, Walters et al. (2005) also describes marked intra- accession variability in storage longevity among the species examined. This leaves the possibility that intraspecific seed mass variation could be one factor that modulates longevity in ex situ storage.

Experimentally Demonstrating Differential Viability Loss of Variable Mass Seeds In Situ and Ex Situ

To test the aforementioned hypotheses, I will obtain and separate fresh

Rudbeckia mollis seeds into three mass-based size classes with an air density separator. Classes will be called light, intermediate, and heavy and will differ in mean seed mass but possess similar variability. Once established, I will test each class for germination ability when fresh under a range of simulated seasonal and constant temperatures. Fresh germination testing will reveal any inherent dormancy or quality

differences between seeds of different mass. To determine inherent physical and physiological differences between classes, I will quantify relative proportions of proteins, lipids, and starch with cellular surface area measurements. I will also perform a developmental study and a seed mass distribution study to understand if seeds of different mass are mature at shedding and to identify where different mass seeds develop on mother plants. I will subsequently store seeds from each mass class in situ and ex situ for two years. In situ storage will consist of seed burial within an established wildflower plot. Ex situ seed storage will consist of four treatments with differing temperature and RH conditions including room temperature (23oC, 42.5% RH), a non- climate controlled shed (variable temperature and RH), refrigerator (4oC, 15% RH), and freezer (-18oC, 15% RH). Every month for six months followed by every two months for a total duration of two years, I will remove and germinate seeds from ex situ and in situ storage treatments to assess mass-dependent viability loss. Finally, I will subject seeds to SSAA at six-month intervals for two years to understand the extent to which high temperature aging is predictive of comparative mass-dependent longevity.

Table 1-1. Overall Cox regression model for the germination response of Rudbeckia mollis seeds of different mass to increasing saturated salt accelerated aging duration. 2 Covariate (xi) Coefficient (i) SE of i Wald  P Hazard ratio Mass class -a - 14.86 0.0006 - Duration -0.17 0.0068 652.81 <0.0001 0.841 Mass class × Days 0.020 0.0075 6.94 0.0084 1.020 Duration × Days 0.0052 0.00079 43.51 <0.0001 1.005 aCoefficients are not calculable for class variables.

Table 1-2. Linear contrasts for germination of Rudbeckia mollis mass classes to increasing duration of saturated salt accelerated aging. Contrasts compare mass classes while controlling for duration. Confidence limits excluding one are significantly different. SSAA durationa Comparison Point estimate 95% confidence limits (days) 0 Light v. Intermediate 0.826 (0.615,1.108) Light v. Heavy 0.901 (0.672,1.209) Intermediate v. Heavy 1.092 (0.814,1.465) 5 Light v. Intermediate 1.162 (0.869,1.554) Light v. Heavy 1.241 (0.927,1.661) Intermediate v. Heavy 1.068 (0.798,1.429) 10 Light v. Intermediate 1.397 (1.043,1.872) Light v. Heavy 1.356 (1.013,1.815) Intermediate v. Heavy 0.970 (0.722,1.303) 15 Light v. Intermediate 1.129 (0.792,1.608) Light v. Heavy 0.970 (0.688,1.369) Intermediate v. Heavy 0.860 (0.607,1.218) 20 Light v. Intermediate 2.008 (1.333,3.026) Light v. Heavy 2.239 (1.470,3.411) Intermediate v. Heavy 1.115 (0.700,1.776) 25 Light v. Intermediate 1.446 (0.781,2.677) Light v. Heavy 1.348 (0.743,2.448) Intermediate v. Heavy 0.932 (0.485,1.794) aThe 30 day SSAA treatment was omitted due to no germination occurring in the intermediate or heavy mass classes.

Figure 1-1. Kaplan-Meier survival functions for Rudbeckia mollis mass classes following saturated salt accelerated aging. A) 0 days, B) 5 days, C) 10 days, D) 15 days, E) 20 days, F) 25 days, and G) 30 days. Mass classes are represented by solid (light), short-dashed (intermediate), and dotted (heavy) lines. Black circles represent censored observations.

Figure 1-2. Germination parameters for Rudbeckia mollis mass-classes following 0-30 days of saturated salt accelerated aging (SSAA). A) Total number of germinated seeds and B) germination rates across mass classes and SSAA durations. Lines in (B) are provided as a visual aid.

CHAPTER 2 CHARACTERIZING GERMINATION ECOLOGY, DEVELOPMENT, DISTRIBUTION, AND INTRACELLULAR ANATOMY IN RUDBECKIA MOLLIS (ASTERACEAE) SEEDS OF DIFFERENT MASS

Introduction

Seed mass is a plant functional trait that varies widely in Spermatophytes. For example, seed mass of a given species will vary between individuals and within individuals of a population. Seed mass varies most within compound inflorescences and is determined by position during development (Vaughton and Ramsey, 1997; Winn,

1991). Seed mass typically decreases acropetally due to resource competition or inflorescence architecture (Diggle, 1995; Stephenson, 1981). In some species, both phenomena may play a role in dictating seed mass (Kilber and Eckert, 2004; Torices and Mendez, 2010). Despite the cause, the downstream consequences of seed mass variation on germination ecology and population level dynamics cannot be understated.

For instance, within a population, higher mass seeds germinate faster and to a greater percentage in most species (Baskin and Baskin, 2014). Across species, higher mass seeds tend to outcompete lower mass seeds in direct competitive scenarios

(Turnbull et al., 1999; Turnbull et al., 2004) or when seeds are covered with leaf litter

(Lönnberg and Eriksson, 2013). Similarly, seedlings from higher mass seeded species display greater survivorship during early seedling establishment or when germination occurs under some particular hazard like shade or herbivory (Moles and Westoby,

2004; Westoby et al., 1996). Seeds of different mass can also exhibit inherent differences in germination ecology due to maternal plant effects. For example, seed position can dictate dormancy in fresh seeds and may lead to season-specific germination for seeds of different mass (Wang et al., 2010). Finally, seeds developing

on different areas of the mother plant can dehisce with differing chemistry. For example, oil concentration increases with seed size in sunflower and is linked to seed position within inflorescences (Hassan et al., 2011; Kaleem et al., 2011; Munshi et al., 2003).

Similarly, mineral nutrient concentration often increases with seed mass including important drivers of seed vigor like nitrogen and phosphorous (Krannitz, 1997; Obeso,

2012). For instance, Peucedanum oreoselinum seeds from a roadside population had greater mass, nitrogen, and phosphorous content and germinated faster and to a higher percentage than seeds from a forest population with less mass, nitrogen, and phosphorous content (Kolodziejek, 2017).

The purpose of this research was to characterize the spatial distribution of different mass seeds within Rudbeckia mollis L. (Asteraceae) plants with the goal of ultimately relating potential differential seed anatomy and physiology to developmental location. I chose R. mollis because previous research characterized this species germination biology on a population and seed mass basis (Kettner and Pérez, 2012;

Genna and Pérez, 2016). In this paper, I examine germination capacity, seed development within inflorescences, pinpoint the origin of seed mass variation within plants, and explore anatomical and physical characteristics of different mass seeds.

This study provides a foundation for subsequent research to target specific regions in R. mollis that 1) represent the source of the greatest variation in seed mass and 2) may be potentially driving intra-population ecological differences.

Materials and Methods

Plant Material and Establishing Mass Classes

Rudbeckia mollis is native to the southeastern United States. Current literature describes R. mollis as an annual, biennial, or short-lived perennial adapted to dry sandy

soils in sandhills and open hammocks (Norcini and Aldrich, 2007; Wunderlin and

Hansen, 2011). Rudbeckia mollis blooms during summer with large radiate head inflorescences consisting of yellow ray florets and dark purple disc florets. Rudbeckia mollis produced achenes, referred to hereafter as seeds, that may be non-dormant at shedding or possess non-deep physiological dormancy alleviated by dry after-ripening

(Kettner and Pérez, 2012).

Fresh R. mollis seeds were hand collected from approximately 25 plants growing on a wildflower seed farm (Wildflowers of Florida Inc., Alachua, FL) on August 23, 2016 when naturally dehiscent. The lot was fractionated with an air density separator into 21 fractions. I removed the nine lowest mass fractions due to insufficient seed fill. I assessed the mean and variance of the 12 remaining fractions after weighing four random samples of 25 seeds each. A one-way analysis of variance (ANOVA) and

Bonferroni’s method provided statistical separation between fractions. I combined fractions with similar mean seed mass to create three mass based classes called light, intermediate, and heavy. Mean seed mass was 383 ± 7, 409 ± 8, 447 ± 10 µg (mean ±

SE) in the light, intermediate, and heavy classes, respectively after resampling each class with four random samples of 50 seeds. Each class was statistically different (F2,9 =

54.95, P <0.0001).

All subsequent experimentation will utilize these classes except for calculating true density. I measured seed density with a separate population due insufficient seed quantity in the original population. Seeds from this population were collected from

Keystone Heights, Florida (25% of seeds 29°43’47.6”N 81°57’20.0”W, 75% of seeds

29°44’44.9”N 81°58’56.1”W) on August 17, 2017. I created three mass classes as

described above and mean seed mass was 420 ± 9, 456 ± 8, 502 ± 6 µg (mean ± SE) in the light, intermediate, and heavy classes, respectively. Each class was statistically different (F2,9 = 112.5, P <0.0001).

Germination of Fresh Seeds

Nine hundred seeds from each mass class were subjected to a germination test under a range of seasonal alternating and constant temperatures on September 7,

2016. Seeds were exposed to a 12 hour photoperiod inside incubators (photosynthetic photon flux density, 52 ± 5 μmol m-2 s-1). The seasonal alternating temperatures included simulated winter (22/11oC), late fall/early spring (27/15oC), early fall/late spring

(29/19oC), and summer (33/24oC). Constant temperatures included 15, 20, 25, 30, and

35oC. Seeds were placed inside plastic germination boxes on blue blotter paper hydrated with autoclaved (117.7 kPa, 121 °C, 40 minutes) deionized water. I used four subsamples of 25 seeds per mass class and temperature combination. Non-germinated seeds that remained after 28 days were tested for viability with the tetrazolium staining assay (data not shown).

Greenhouse Study

Plant establishment

Three seeds from each mass class were sown in 15, 0.4 L pots (Nursery

Supplies, PA, USA) filled with Sunshine #4 media (Sun Gro Horticulture, MA, USA) during August 2017. About twenty percent of seeds germinated and remained as basal rosettes until February 2018 when more plantlets emerged. Nine plants from each mass class were selected and transferred to 2.78 L pots filled with Sunshine #4 media on

February 21, 2018. Pots were placed in a 30% shadehouse and hand watered as needed. I removed weeds and other plants inconsistent with R. mollis phenotype

leaving six plants originating from light seeds and eight plants each originating from intermediate and heavy seeds. I transferred all plants to 11.36 L pots on June 14, 2018 filled with Sunshine #4 media. Slow release fertilizer (15-9-12, N-P-K, Osmocote, The

Scotts Company, OH, USA) was added to pots at this time. I separated all 22 plants into two groups of 11 plants for 1) a seed developmental study and 2) a seed mass distribution study. I randomized all plants on four benches within the greenhouse. Mass class origin was not a factor in the developmental study but was a factor in the seed mass distribution study. Therefore, in the developmental study, I randomly tagged inflorescences among all plants upon anthesis. However, in the seed mass distribution study, information about seed mass origin was conserved and factored into subsequent analyses.

Developmental study

My objective here was to understand if seed maturity at harvest was similar among seeds in different locations within inflorescences. Therefore, I targeted lower, middle, and upper inflorescence regions with the assumption that seed mass will vary among sampled fruits. Rudbeckia mollis inflorescences are indeterminate and take approximately two to four weeks to complete anthesis. I tagged inflorescences when basal florets opened and started the developmental study when the uppermost distal floret opened. I harvested inflorescences for ten weeks thereafter.

I placed all inflorescences in a plastic box with a tight fitting lid to minimize water loss during transfer to the laboratory. I sampled fruits following a single spiral starting with basal florets and ending with upper florets (Figure 2-1). I maintained sampled fruits in the order of harvest and divided the total number of fruits by three. I then placed fruits in a lower, middle, and upper group. I placed extra fruits in the lower group if the total

number of fruits was not divisible by three. I measured fresh and dry mass of lower, middle, and upper groups gravimetrically. Dry mass was determined after drying samples in a forced air oven at 90oC oven for two days. I calculated water content on a dry mass basis. Data is presented on a single seed basis to account for different group sizes.

Seed mass distribution study

Plants senesced quickly in the fall of 2018 and inflorescences were harvest over the course of a week beginning on September 25, 2018. I measured plant height and stem diameter 2.5 cm above the soil surface and counted the total number of inflorescences and primary branches originating from the main stem prior to harvest.

Next, I chose primary branches originating from the main stem to collect inflorescences.

I always sampled the lowermost branch and the uppermost branch. I randomly sampled at least five branches per plant. For each branch, I harvested and labelled each inflorescence to indicate its position and placed inflorescences upright to prevent seed shedding. All inflorescences acclimated for one month in the laboratory before sampling.

I analyzed each inflorescence by first measuring width and height with calipers.

Next, I removed individual seeds following a single spiral to obtain individual seed masses (see Figure 2-1). Most inflorescences had unfilled seeds throughout the inflorescence with the highest empty seed concentration in the uppermost portion. In some cases, florets were also not open in the upper region. Unfilled seeds were identifiable by either their light brown color or by a press test and excluded from analyses. The number of seeds sampled per inflorescence varied considerably

(minimum = 2, maximum = 26, median = 9, skewness = 0.632, kurtosis = 0.207). The

total number of seeds sampled from each class also differed since the number of plants sampled per class and the number of inflorescences sampled per plant varied (light =

669 seeds, intermediate = 1029 seeds, heavy = 1079 seeds).

Comparative Seed Anatomy

I selected seven seeds of different mass for embedding and microscopy. Ultra- thin sections and transmission electron microscopy (TEM) images were taken from seeds weighing 380.8, 389.1, 405.4, 427.8, 433.8, 446.9, and 466.2 µg. I also obtained semi-thick sections to image with a compound light microscope.

To prepare seeds for microscopy, pericarp tissue was removed to expose the embryo since preliminary attempts found that the pericarp hindered resin infiltration.

Seeds were soften in water for 15 minutes, dried, and fixed overnight in 4% paraformaldehyde and 2% glutaraldehyde in 0.1M cacodylate buffer with 2 mM MgCl,

1mM CaCl, and 0.25% NaCl (pH 7.21). Fixed samples were processed with the aid of a

Pelco BioWave Pro laboratory microwave (Ted Pella, Redding, CA, USA). Seeds were encapsulated in 3% agarose, washed in 0.1M cacodylate buffer with additives (pH

7.24), and post-fixed in buffered 2% OsO4 overnight at 4°C. Samples were buffer washed, water washed, and dehydrated in a graded ethanol series from 25% through

65% with 5% increments. Seeds were processed in 75% ethanol containing 1% p-

Phenylenediamine before dehydrating in 75% ethanol. Dehydration in graded ethanol series was continued in 75% through 100% with 5% increments followed by triple 100% anhydrous acetone washes. Dehydrated specimens were infiltrated in 30%, 40%, 50%,

60%, 70% and 100% acetone/Spurr’s epoxy resin with Z6040 embedding primer, and cured at 60°C for 48 hours. Ultra-thin sections were collected on carbon coated Formvar

slot copper grids, post-stained with 2% aqueous uranyl acetate and Reynold’s lead citrate.

Ultra-thin sections were examined with a FEI Tecnai G2 Spirit Twin TEM (FEI

Corp., Hillsboro, OR). Digital images were acquired with a Gatan UltraScan 2k x 2k camera and Digital Micrograph software (Gatan Inc., Pleasanton, CA) and an AMT-

XR41 1k x 1k camera and TIA software (FEI Corp., Hillsboro, OR). For each seed, I imaged five cells in the meristematic regions of the radicle and apical meristem. I also imaged five cells in the middle of one cotyledon. I used Image J (U.S. National Institutes of Health, Bethesda, Maryland, USA) to measure cell width as the average of perpendicular intersecting widths. I also measured total cell, nucleus, protein body, and starch granule surface area (SA). I calculated lipid SA by subtracting nucleus and protein body SA from total cell SA.

Seed Density

I calculated seed density with the toluene displacement method and a 10 mL pycnometer (Wilmad-LabGlass, NJ, USA) similar to Razavi and Milani (2006). Briefly, I weighed four replications of approximately 2 grams of seeds from each mass class. I added each sample to the pycnometer, filled the pycnometer with toluene, and reweighed the pycnometer. Seed volume was determined by dividing the mass of toluene displaced by the density of toluene. Seed density was determined by dividing the mass of each sample by volume determined with the pycnometer.

Statistical Analyses

Germination testing

I used non- and semi-parametric time-to-event analyses to analyze germination data. Germination was the event of interest and coded as 1. Moldy seeds removed

before 28 days and non-germinating seeds after 28 days were censored and coded as

0. I generated Kaplan-Meier survival functions by stratifying data by temperature and mass class. Next, I built Cox regression models for the seasonal alternating and constant temperature data sets. In both cases, I treated mass class as a class variable and temperature as a continuous variable. I verified the proportional hazards assumption with graphical and residual analyses. In both data sets, a temperature*day interaction was necessarily included to satisfy the proportional hazards assumption for temperature. I used orthogonal contrasts to compare temperature treatments and mass classes within temperature treatments.

Seed mass distribution

I assessed the distribution of seed mass within R. mollis plants with a four level hierarchical ANOVA model. The population of plants in this study was the highest order classification (termed population). All plants were nested within the mass class that seeds originated from (termed mass class). Plants were coded as 1, 2, and 3 in the light, intermediate, and heavy mass classes, respectfully. Therefore, if a plant originated from a light class seed, the plant was classified as light. Primary branches originating from the main stem were nested within mass classes (termed primary branch). The lowermost primary branch was coded as 1 with higher primary branches coded sequentially thereafter. Secondary branches originate from primary branches and terminate into inflorescences. Secondary branches were nested within primary branches (termed secondary branch). The first secondary branch encountered, located closest to the primary branch insertion site, was coded 1 and higher secondary branches were coded sequentially thereafter. Finally, I nested individual seeds within secondary branches (termed within inflorescences (error)). The first seed located

basally in each inflorescence was coded as 1 and each distal seed sampled was coded sequentially thereafter. Seed number varied between inflorescences resulting in an unbalanced design. Therefore, F-tests are not reported.

Furthermore, I performed two correlation analyses. The first analysis included mean inflorescence seed mass, mass class, primary branch position, secondary branch position, inflorescence width, and inflorescence height. The second analysis included individual seed mass, mass class, primary branch position, secondary branch position, and seed position within inflorescences. Each analysis included both ordinal and continuous data. Therefore, I reported Spearman rank correlation coefficients.

Developmental study, comparative seed anatomy, true density

All developmental study data, plant characteristics, and density measurements were compared with a one-way ANVOA checking assumptions graphically. I used the nonparametric Kruskal-Wallis ANOVA if normality was violated. I separated means with the Bonferroni method at α = 0.05.

Results

Germination Testing When Fresh

Fresh R. mollis seed germination ranged from 68-94% in all classes and temperatures excluding summer (33/24oC) and constant 35.0oC. Temporal patterns of germination for seeds exposed to simulated seasonal temperatures were distinct between mass classes in response to winter (22/11oC) and late fall/early spring

(27/15oC) temperatures but homogenous in response to early fall/late spring (29/29oC) temperatures. The probability of not germinating was ≥ 0.93 for all mass classes in response to simulated summer conditions (Figure 2-2). Cox regression modeling showed that the temperature main effect was highly significant and negative for R.

mollis germination response to seasonal alternating temperatures indicating that the probability of germination decreased with increasing temperature across mass classes.

A temperature*day covariate was also significant and indicates that the probability of germination decreased with duration of exposure (Table 2-1). Orthogonal contrasts revealed that all seasonal temperature comparisons were significant except when

22/11oC was compared to 27/15oC (Table 2-2). Furthermore, contrasts showed that R. mollis germinates optimally at 29/19oC when controlling for seed mass. The mass class main effect was highly significant indicating that seed mass influences germination response to seasonal temperatures (Table 2-1, P <0.0001). For both 22/11 and 27/15oC treatments, the probability of germination for light and intermediate seeds was 1.5 to 1.9 times higher than heavy class seeds whereas light and intermediate seeds germinated similarly (Table 2-3).

Constant temperatures elicited a different response in fresh R. mollis seeds. The probability of germination increased from 15 to 25oC and decreased from 25 to 35oC when controlling for seed mass (Figure 2-3, Table 2-2). Therefore, fresh R. mollis seeds germinate optimally at 25.0oC. The main effect of constant temperature was not significant when an interaction with time was necessarily included in the model.

However, the temperature*day covariate was highly significant and indicates that the probability of germination decreased during the germination study. Unlike seasonal temperatures, seed mass did not significantly influence germination response in R. mollis mass classes following exposure to constant temperatures (Table 2-1, P = 0.14).

Developmental Study

Anthesis ranged from 1 to 5 weeks (mean = 3.3) in all inflorescences sampled.

Inflorescences are relatively flat when basal florets open (Figure 2-4 A-B). However,

upon the conclusion of anthesis, inflorescence height increases to create a conical shape and ray petals unfurl and increase in size (Figure 2-4C). From 4-6 weeks following anthesis, no discernable change in inflorescence structure is detectable except for ray florets beginning to brown (Figure 2-4 D-F). From 7-8 weeks post anthesis, inflorescences initiate senescence, turn brown, and dry (Figure 2-4 G-H).

Seeds easily dehisced with wind or by incidental contact around week eight after anthesis (personal observation). At weeks 9-10 following anthesis, inflorescences are fully senesced and disk florets fall out as inflorescences dry and open (Figure 2-4 I-J).

Seed fresh mass declined from weeks 4-7 in seeds sampled from lower, middle, and upper regions of R. mollis inflorescences. Fresh mass was significantly different at weeks 7 and 10 (P value range 0.02-0.004). Dry mass remained relatively constant from weeks 4-10 but was significantly different at weeks 7, 9, and 10 (P value range 0.02-

0.006). Seeds collected from upper inflorescence regions displayed reduced fresh and dry mass than lower regions when significant differences were detected. Water content declined in all regions from weeks 4-7 but most prominently in the upper region. Water content was lowest at week 7 and fluctuated thereafter with rain and relative humidity

(Figure 2-5).

Seed Mass Distribution

Rudbeckia mollis branches alternate spirally around a main stem. Primary branches extend from the main stem and secondary branches extend from primary branches. Each branch terminates into an inflorescence including the terminal portion of the main stem (Figure 2-6). Overall, plants originating from light, intermediate, and heavy seeds did not differ with respect to plant height, stem diameter, primary branch number, inflorescence number, or mean inflorescence width. However, mean

inflorescence height was greatest in light class plants at 14.0 ± 0.08 mm and similar among intermediate and heavy plants with an average of 12.9 ± 0.66 mm. For all seeds collected from plants originating from light, intermediate, or heavy class seeds, mean seed mass was lowest in plants originating from heavy seeds whereas plants originating from light or intermediate seeds did not differ with respect to mean seed mass (F2,8 =

12.17, P = 0.0037). Furthermore, seed mass distribution differed with respect to seed origin despite a similar range of seed mass across classes (mean range = 496 µg). Both minimum and maximum seed mass decreased by an average of 17 and 8%, respectfully, from light to heavy class plants (Table 2-4).

In agreement with descriptive statistics in Table 2-4 and Figure 2-6, 31.1% of variation in seed mass occurred between plants within the population studied. However, most seed mass variation occurred within inflorescences, accounting for 40.3% of variation. The smallest source of variation was found in primary branches, with 3.5 times more variation occurring in secondary branches. Only 5.3% of variation in seed mass existed between plants within mass classes (Table 2-5).

Two correlation analyses further characterize R. mollis plant features. The first correlation analysis utilized mean inflorescence seed mass and information about primary or secondary branch location, inflorescence width and height, and the mass class that seeds originated from (Table 2-6). Firstly, mean inflorescence seed mass displayed a moderately negative correlation with mass class indicating that seed mass was lower in heavy class plants (r = -0.59, P = <0.0001). Next, inflorescence height was weakly negatively correlated with mass class indicating that inflorescence height was highest in light class plants and decreased from intermediate to heavy class plants (r = -

0.13, P = 0.023). Primary branch number and secondary branch number were weakly negatively correlated indicating that higher order primary branches contained fewer secondary branches (r = 0.12, P = 0.038). For both primary and secondary branches, inflorescence width was weakly positively correlated indicating that inflorescence width increased in higher order primary branches and secondary branches (r range 0.20-0.18,

P range 0.0005-0.0015). Inflorescence height was also weakly positively correlated to mean inflorescence seed mass indicating that taller inflorescences contained higher mass seeds on average (r = 0.19, P = 0.0013). The strongest association was found between inflorescence width and height, which were both strongly positively correlated

(r = 0.66, P = <0.0001).

The second correlation analysis utilized individual seed masses and information about seed position within each inflorescence, location of primary and secondary branches, and the mass class that seeds originated from (Table 2-7). First, seed position within inflorescences showed a weak positive correlation to mass class indicating that the number of filled seeds was lowest in light class plants and highest in heavy class plants (r = 0.087, P = <0.0001). Again, individual seed mass was moderately negatively correlated to mass class, which is a similar result when mean inflorescence seed mass was correlated with mass class (r = -0.51, P = <0.0001, Table

2-5). A second similar conclusion to Table 2-6 is found when primary branch number is correlated to secondary branch number, resulting in a weak negative association (r = -

0.17, P = <0.0001). This indicates that higher order primary branches had fewer secondary branches. There was a weak positive association between seed position within inflorescences and primary or secondary branches (r range 0.040-0.056, P range

0.033-0.037). This indicates that the number of filled seeds sampled was highest in higher order primary and secondary branches. There were weak correlations between individual seed mass and primary or secondary branch positions. However, seed mass in primary branches increased from lower to upper branches (r = 0.13, P = <0.0001) while seed mass in secondary branches decreased from proximal to distal locations (r =

-0.048, P = 0.011). Finally, a weak negative association between seed mass and seed position within inflorescences indicates that seed mass declined from basal to apical fruits (r = -0.20, P = <0.0001).

Comparative Seed Anatomy and Density

Light microscopy revealed similar cellular organization in R. mollis seeds of different mass (data not shown). Total cell and protein body SA in meristematic regions of the radicle and apical meristem appeared smaller compared to cells in the cotyledon

(Figure 2-7 A-C). TEM images showed very little intracellular free space across seeds of different mass and regions within embryos. TEM images also revealed differing cell anatomy in each region. Cells in the radicle and apical meristem had numerous protein bodies and a visible nuclear region. Cells in the cotyledon also had numerous protein bodies but fewer visible nuclei. Protein bodies in the cotyledon were also filled with numerous distinct starch granules whereas protein bodies in the radical and apical meristem had fewer starch granules (Figure 2-7 D-F).

There was no clear change in any dimensional category in the radical, apical meristem, or cotyledon regions across seeds of different mass. However, cell width, cell

SA, protein SA, starch SA, and lipid SA was lowest in the radicle region and increased from the apical meristem to cotyledon regions. Only nucleus SA appeared lowest in the cotyledon region in comparison to the radicle and apical meristem. The single greatest

contributor to SA in each region was the lipid fraction with an average of 67 ± 2.6%

(mean ± SE) across different mass seeds and regions (Table 2-7).

Finally, seed density decreased by 1.2% with increasing seed mass. Seed density was 1.085 ± 0.018, 1.074 ± 0.012, and 1.072 ± 0.014 in the light, intermediate, and heavy classes, respectfully. However, differences were not significant (F2,9 = 0.90,

P = 0.45).

Discussion

Germination Testing When Fresh

Rudbeckia mollis seeds in this study germinated similarly to naturally collected stands reported by Kettner and Pérez (2012) when controlling for seed mass. In this research and in Kettner and Pérez (2012), seeds were less than one month old and germination was markedly reduced in simulated summer conditions and at constant

35.0oC. This is likely a mechanism to avoid germination during a period of warmer weather during the summer or early fall following dehiscence. However, following a period of storage, R. mollis germinates to an average of 41 and 28% in response to simulated summer and 35.0oC, respectfully (Genna and Pérez, 2016). Therefore, we can reasonably conclude that R. mollis displays non-deep physiological dormancy at shedding (Baskin and Baskin, 2014).

Further, germination capacity is dependent on mass when fresh R. mollis seeds are separated into distinct mass classes. In this research, heavy class seeds germinated to 73% under simulated winter (22/11oC) and late fall/early spring (27/15oC) temperatures while light and intermediate class seeds germinated to 91%. Therefore, under conditions that seeds are likely to experience following dehiscence in the fall, lower mass seeds are nondormant and higher mass seeds are conditionally dormant.

Similar mass dependent germination was reported in Eremopyrum distans (Poaceae) whereby higher mass seeds located basally in inflorescences are shed with higher non- deep physiological dormancy compared to lighter seeds located distally (Wang et al.,

2010). Differential germination response at shedding may be a bet-hedging strategy to spread germination events and the associated risk through time: a common feature of annual plants occurring in high-risk environments (Gremer and Venable, 2014;

Tielbörger et al., 2012). Rudbeckia mollis is known to emerge following shedding in the fall if rain is sufficient. Fall germinating plants remain as basal rosettes until the following spring when temperature and soil moisture are favorable for rapid growth. This research suggests that the small percentage of fall germinating seeds may be of lower mass and empirical research is needed to demonstrate this.

As previously stated, differential physiological dormancy is lost after a period of time in R. mollis and seeds of all masses germinate similarly at all alternating seasonal temperatures (Genna and Pérez, 2016). Similar germination in non-dormant R. mollis seeds is in contrast, however, with numerous reports that higher mass seeds germinate to a greater extent and/or at a faster rate compared to lower mass seeds within a population (Afonso et al., 2014; Kolodziejek, 2017; Moravcova et al., 2005; Torices et al., 2013). However, similar germination in different mass seeds that are presumed non- dormant have been reported in other species (Guillemin and Chauvel, 2011; Hernandez and Orioli, 1985; Parker et al., 2006; Torices and Mendez, 2010). It is not currently known if R. mollis can form a persistent seed bank or if seeds are capable of surviving burial in soil (Chapter 3). Successful establishment and recruitment of new individuals the season following shedding appears to be critical in this summer annual species.

Developmental Study and Distribution of Seed Mass

Seed development occurs rapidly in R. mollis inflorescences following pollination and fertilization and seeds in all regions of the inflorescence are mature by week 7 and only differ with respect to mass. Rudbeckia mollis plants grown in this research also differed somewhat from plants observed in the wild. For instance, plants appeared shorter, branched more, and had more inflorescences than wild plants (personal observation). The most surprising finding here is that average mass of all seeds collected from plants started from heavy class seeds was lower compared to the seed crop collected from plants originating from light class seeds. However, the limited number of plants used in this study precludes broad conclusions about the heritability of seed mass in R. mollis and may be a product of phenotypic plasticity. Repeating this research with more plants is necessary to draw any broad conclusions about wild R. mollis plants.

In R. mollis, 68.8% of seed mass variation occurred within plants and 30.1% of seed mass variation occurred between individuals in the population. A greater within- plant seed mass variance compared to among plant variance has been demonstrated repeatedly in the literature (Hendrix and Sun, 1989; Michaels et al., 1988; Obeso, 1993;

Vaughton and Ramsey, 1997; Winn, 1991). This research was largely conducted in response to the supposition that seed mass variation was low within species (Harper et al., 1970) and that theoretical models predicted an optimum offspring size for a given species in a given environment (McGinley et al., 1987; Smith and Fretwell, 1974). The greatest component of within plant seed mass variation occurred within R. mollis inflorescences and accounted for 40.3%. For other species that produce inflorescences, this finding is also in agreement and suggests that most within plant seed mass

variation originates from the inflorescence (Hendrix and Sun, 1989; Vaughton and

Ramsey, 1997; Winn, 1991). Regarding the distribution of seed mass within individual plants (i.e. where do seeds of different mass come from), seed mass was greatest in higher order primary branches, lower order secondary branches, and proximal positions within inflorescences. In contrast to R. mollis, seed mass was greatest in higher order secondary branch positions in Prunella vulgaris (Winn, 1991) and seed mass was lowest in higher order primary branches in Asphodelus albus (Obeso, 1993).

Discrepancies in seed mass distribution between these species may be due to differences in plant structure.

Many seeds were unfilled or florets did not appear open in distal inflorescence regions, and the number of filled seeds per inflorescence was greatest in higher order primary and secondary branch positions (i.e. high and towards the outside on the plant).

Empty seeds were most likely aborted at some time following pollination: a common feature of the Asteraceae. In sunflower, reducing competition for resources within inflorescences increases the number of filled seeds (Alkio et al., 2003). We do not know what controls the decline in seed mass within R. mollis inflorescences and in particular if the acropetal decline is due to resource or architecture effects or a combination of both.

However, in this research, I suspect that many florets that appeared unopened were likely not pollinated. Rudbeckia mollis inflorescences are indeterminate and pollinators frequently visited lower florets during anthesis and were conspicuously absent when upper florets were open (personal observation). These findings suggest that upper florets are more likely to self-pollinate than outcross and reduced seed set in upper

florets could be due to non-uniform pollination (Dai et al., 2018). A manual pollination study is necessary to address these findings.

Comparative Seed Anatomy and Density

This research demonstrated that R. mollis seeds of different mass are anatomically similar with respect to all dimensional categories and density. Across regions within embryos, all dimensional categories were highest in the cotyledon region except for nucleus SA. Our data also show that starch, protein, and lipid SA change proportionally with cell width and total cell SA in all regions of the embryo. For example, the protein, starch, and lipid fraction in the cotyledon had a narrow range of 26-30, 3.5-

5.0, and 70-74%, respectfully, across seeds of different mass. However, total protein, starch, and lipid SA had a considerably higher range of 41-114, 6.2-16, 94-452 µm2, respectfully, across seeds of different mass. In soybean, protein, starch, soluble sugars, and lipids increased with mass (Guleria et al., 2008). Similarly, protein, micronutrients, and macronutrients increased with seed mass in wheat (Bramble et al., 2002; Calderini and Ortiz-Monasterio, 2003) but nitrogen concentration did not in a different study

(Stoddard, 1999). In wild species including Hedra helix (Araliaceae) and Purshia tridentate (Rosaceae), carbon, nitrogen, phosphorous, and sulfur increased with seed mass (Krannitz, 1997; Obeso 2012).

Furthermore, lipid concentration quantified with TEM did not decline with decreasing seed mass in R. mollis as is described in sunflower inflorescences (Hassan et al., 2011; Kaleem et al., 2011; Munshi et al., 2003). However, SA calculations collected in this research may not align with concentration determined with gas chromatography. For example, average protein and lipid components were 20 and 67%, respectfully, across the radicle, apical meristem, and cotyledon regions. Published

components of closely related Rudbeckia hirta had protein and lipid fractions of 29 and

30%, respectfully (Seed Information Database, Royal Botanic Gardens, Kew).

Therefore, different mass seeds in different species largely have differing chemistry, but

SA calculations in this study were insufficient to reveal differences in R. mollis seeds of different mass. SA measurements may also have considerably overestimated the lipid fraction.

Conclusion

This research revealed that R. mollis seeds of different mass germinate differently when fresh, develop in different locations within mother plants, and have similar cellular anatomy. More specifically, higher mass R. mollis seeds display the greatest conditional dormancy when fresh, higher mass seeds are located closest to resources from the main stem and within inflorescences, and the greatest variation in seed mass occurs within inflorescences. Therefore, the inflorescence is an interesting location to study differential physiology at shedding and the interaction between position, mass, and dormancy. For example, does seed position and possible resource effects drive differential dormancy at shedding? How may resource competition within inflorescences control hormone expression in developing seeds? This can be studied by isolating position and resource driven effects through selective floret removal and monitoring changes in primary dormancy. For example, higher mass seeds are the greatest resource sinks within inflorescences and also have the greatest conditional dormancy at shedding. If proximal florets are removed during development, do distal seeds both gain mass and express greater conditional dormancy? Rudbeckia mollis is a wild species that can function as a model for this research and can drive understanding of differing ecological roles for different mass seeds within populations.

Table 2-1. Cox regression models for the germination response of fresh Rudbeckia mollis seeds of different mass exposed to simulated seasonal alternating and constant temperatures. Experiment Covariate (xi) Coefficient SE of Wald 2 P Hazard (βi) βi Ratio Seasonal Temperature -0.047 0.015 9.46 0.0021 0.954 Temp*days -0.014 0.0020 48.3 <0.0001 0.986 Mass class -a - 19.6 <0.0001 - Constant Temperature -0.00060 0.0085 0.0052 0.94 0.999 Temp*days -0.012 0.0013 77.8 <0.0001 0.988 Mass class - a - 3.93 0.14 - aCoefficients are not calculable for class variables.

Table 2-2. Orthogonal contrast comparisons for germination of Rudbeckia mollis mass classes to seasonal and constant temperatures controlling for seed mass. Confidence limits excluding one are significant at α = 0.05. Experiment Comparison Hazard Ratio 95% Confidence limits Seasonal 22/11 vs 27/15 1.063 (0.889, 1.272) 22/11 vs 29/19 0.774 (0.650, 0.923) 22/11 vs 33/24 33.86 (20.87, 54.93) 27/15 vs 29/19 0.728 (0.609, 0.871) 27/15 vs 33/24 31.84 (19.65, 51.60) 29/19 vs 33/24 43.72 (26.97, 70.87) Constant 15 vs 20 0.411 (0.345, 0.489) 15 vs 25 0.337 (0.283, 0.400) 15 vs 30 0.976 (0.817, 1.166) 15 vs 35 171.7 (54.93, 536.8) 20 vs 25 0.820 (0.691, 0.973) 20 vs 30 2.377 (1.986, 2.845) 20 vs 35 418.2 (133.7, 1308) 25 vs 30 2.898 (2.422, 3.467) 25 vs 35 510.0 (163.1, 1595) 30 vs 35 175.7 (56.30, 550.0)

Table 2-3. Orthogonal contrast comparisons for germination of Rudbeckia mollis mass classes within seasonal temperatures. Confidence limits excluding one are significant at α = 0.05. Experiment Comparison Hazard Ratio 95% Confidence limits 22/11 Light vs Intermediate 1.131 (0.844, 1.516) Light vs Heavy 1.654 (1.213, 2.255) Intermediate vs Heavy 1.462 (1.073, 1.991) 27/15 Light vs Intermediate 0.889 (0.655, 1.205) Light vs Heavy 1.728 (1.251, 2.385) Intermediate vs Heavy 1.944 (1.411, 2.678) 29/19 Light vs Intermediate 1.159 (0.853, 1.574) Light vs Heavy 1.138 (0.844, 1.534) Intermediate vs Heavy 0.982 (0.723, 1.334) 33/24 Light vs Intermediate 1.187 (0.399, 3.532) Light vs Heavy 1.377 (0.437, 4.338) Intermediate vs Heavy 1.160 (0.354, 3.801)

Table 2-4. Rudbeckia mollis plant characteristics. Plants were started from seeds originating from light (n = 3 plants), intermediate (n = 4 plants), and heavy (n = 4 plants) mass classes. Seed mass data was determined from a total of 669 seeds in the light class, 1029 seeds in the intermediate class, and 1079 seeds in the heavy class. Plants were grown in a 30% shadehouse, fertilized with a 15-9-12 slow release fertilizer, and hand watered as needed. Different letters indicate significant differences following means separation with the Bonferroni method at α = 0.05.

Seed mass (µg)

ranch

Original seed mass Plantheight ± SE (cm) Stem diameter ± SE (mm) b Primary numberSE ± Inflorescence numberSE ± Mean inflorescence ± SEwidth (mm) Mean inflorescence ± SEheight (mm) Minimum Maximum Range Mean ± SE 67 ± 7.9 ± 14 ± 107 ± 13.2 ± 14.0 ± 542 ± Light 296 787 491 24 a 2.3 a 7.5 a 52 a 0.87 a 0.08 a 34 a 97 ± 9.8 ± 14 ± 115 ± 13.7 ± 12.9 ± 499 ± Intermediate 260 776 516 16 a 1.6 a 6.3 a 41 a 0.85 a 0.39 a 33 a 71 ± 8.7 ± 16 ± 127 ± 12.9 ± 12.8 ± 439 ± Heavy 247 727 481 10 a 1.6 a 8.2 a 46 a 0.82 a 0.92 b 13 b 80 ± 8.9 ± 14 ± 116 ± 13.3 ± 13.2 ± 489 ± mean ± SE 268 763 496 20 1.8 6.6 42 0.83 0.76 50

Table 2-5. Four level hierarchical ANOVA assessing seed mass variation within Rudbeckia mollis plants. Seed number varied between inflorescences resulting in an unbalanced design. F-tests are not reported. Variance Sum of Mean Variance Percent of df source squares Square component total Total 2776 20816160 7497 8479 100 Population 2 5161654 2580827 2640 31.1 Mass class 8 1304233 163029 451 5.3 Primary 51 2298391 45066 443 5.2 branch Secondary 139 3243898 23337 1524 18.0 branch Inflorescences 2576 8807984 3419 3419 40.3 (error)

Table 2-6. Spearman correlation coefficients and probability statistics of mean inflorescence seed mass and measured plant characteristics. Mass class was coded as 1 = light, 2 = intermediate, and 3 = heavy. The lowermost primary branch was coded as 1 and higher branches were coded sequentially thereafter. The first secondary branch within a primary branch was coded as 1 and distal secondary branches were coded sequentially thereafter. P-values are in parentheses. Mass Primary Secondary Inflorescence Inflorescence Mean inflorescence

class branch branch width height seed mass Mass class - -0.025 (0.67) -0.011 (0.85) -0.028 (0.63) -0.13 (0.023) -0.59 (<0.0001) Primary branch - - -0.12 (0.038) 0.20 (0.005) 0.073 (0.21) 0.10 (0.087) Secondary - - - 0.18 (0.0015) 0.10 (0.074) -0.011 (0.85) branch Inflorescence - - - - 0.66 (<0.0001) -0.018 (0.76) width Inflorescence - - - - - 0.19 (0.0013) height Mean inflorescence ------seed mass

Table 2-7. Spearman correlation coefficients and probability statistics of individual seed masses and measured plant characteristics. Mass class was coded as 1 = light, 2 = intermediate, and 3 = heavy. The lowermost primary branch was coded 1 and higher branches were coded sequentially thereafter. The first secondary branch within a primary branch was coded as 1 and distal secondary branches were coded sequentially thereafter. Seed position within inflorescences was coded as 1 for the first proximal seed located in the outermost whorl and sequentially thereafter for distal seeds located in inner whorls. P-values are in parentheses. Seed position within Individual seed Mass class Primary branch Secondary branch inflorescences mass Mass class - -0.033 (0.84) 0.0024 (0.90) 0.087 (<0.0001) -0.51 (<0.0001) Primary branch - - -0.17 (<0.0001) 0.040 (0.037) 0.13 (<0.0001) Secondary branch - - - 0.056 (0.0033) -0.048 (0.011) Seed position within - - - - -0.20 (<0.0001) inflorescences Individual seed mass - - - - -

Table 2-8. Anatomical measurements from different locations within Rudbeckia mollis seeds of different mass. Means for each combination of location, dimension, and seed mass are the average of five cells. Cell width units are µm. Surface area (SA) units are µm2. Means are calculated for each combination of dimension category and location across seeds of different mass. Seed mass (µg) Mean Location Dimension 380.8 389.1 405.4 427.8 433.8 446.9 466.2 ± SE 12.1 9.24 11.7 9.86 12.3 11.5 9.52 10.9 Radicle Cell width ± 2.0 ± 1.5 ± 1.2 ± 2.1 ± 1.5 ± 2.0 ± 0.55 ± 1.9 128 73.3 122 87.4 131 113 76.1 104 Cell SA ± 39 ± 22 ± 29 ± 36 ± 33 ± 34 ± 5.7 ± 36 11.3 28.4 9.56 16.6 26.0 26.7 23.7 20.3 Nucleus SA ± 16 ± 12 ± 14 ± 12 ± 18 ± 20 ± 7.1 ± 15 22.1 2.23 32.0 15.9 21.2 14.0 8.15 16.5 Protein SA ± 12 ± 2.6 ± 13 ± 12 ± 8.7 ± 18 ± 3.1 ± 14 1.57 0.004 0.397 0.084 0.028 0.103 0.627 0.402 Starch SA ± 3.3 ± 0.005 ± 0.78 ± 0.12 ± 0.042 ± 0.23 ± 0.57 ± 1.3 95.0 42.7 80.9 54.9 83.3 72.8 44.2 67.7 Lipid SA ± 30 ± 13 ± 15 ± 21 ± 23 ± 28 ± 5.1 ± 27 Apical 15.1 12.3 14.5 10.3 14.5 13.2 14.3 13.5 Cell width Meristem ± 2.5 ± 1.6 ± 0.74 ± 3.2 ± 1.6 ± 1.6 ± 1.9 ± 2.4 195 125 170 88.5 175 144 157 151 Cell SA ± 63 ± 31 ± 20 ± 40 ± 26 ± 29 ± 34 ± 47 27.7 39.3 18.1 28.6 23.3 26.3 22.7 26.7 Nucleus SA ± 17 ± 14 ± 18 ± 16 ± 6.5 ± 11 ± 3.9 ± 14 27.9 16.9 32.3 2.45 26.9 17.7 33.2 22.5 Protein SA ± 25 ± 9.9 ± 9.8 ± 2.7 ± 15 ± 4.8 ± 17 ± 16 2.17 0.848 6.33 0.031 4.33 1.80 3.13 2.66 Starch SA ± 2.5 ± 0.73 ± 6.3 ± 0.026 ± 4.77 ± 2.2 ± 3.9 ± 3.8 139 68.8 120 57.4 124 100 101 102 Lipid SA ± 44 ± 28 ± 23 ± 28 ± 23 ± 21 ± 21 ± 38

Table 2-8. Continued. Seed mass (µg) Mean Location Dimension 380.8 389.1 405.4 427.8 433.8 446.9 466.2 ± SE 23.4 15.3 17.9 27.2 12.4 13.8 29.3 20.1 Cotyledon Cell width ± 2.8 ± 1.3 ± 2.4 ± 6.4 ± 3.2 ± 2.9 ± 3.0 ± 6.9 448 186 256 650 135 180 732 375 Cell SA ± 73 ± 27 ± 61 ± 374 ± 55 ± 91.5 ± 154 ± 268 3.09 2.30 5.70 6.87 0 1.22 13.8 4.72 Nucleus SA ± 6.9 ± 5.1 ± 8.0 ± 11 ± 0 ± 2.7 ± 8.1 ± 6.8 114 64.2 75.2 191 40.7 51.5 227 110 Protein SA ± 32 ± 22 ± 27 ± 95 ± 20 ± 36 ± 60 ± 81 15.5 6.09 11.5 31.2 6.75 6.22 26.7 14.8 Starch SA ± 8.3 ± 3.0 ± 3.6 ± 18 ± 3.6 ± 6.9 ± 11 ± 13 330 119 175 452 94.4 127 492 260 Lipid SA ± 49 ± 18 ± 30 ± 273 ± 38 ± 54 ± 104 ± 186

Figure 2-1. Example of floret sampling procedure for the seed development and seed mass distribution study. Seeds were harvested beginning with the basal most floret and ending with the distal most floret following a single plane of symmetry. Photographs are courtesy of Nicholas Gerard Genna.

Figure 2-2. Germination response of fresh Rudbeckia mollis mass classes exposed to simulated seasonal alternating temperatures. A) Winter (22/11oC), B) late fall or early spring (27/15oC), C) early fall or late spring (29/19oC), and D) summer (33/24oC). Mass classes are represented by solid (light), short- dashed (intermediate), and dotted (heavy) lines. Black circles represent censored observations.

Figure 2-3. Germination response of fresh Rudbeckia mollis mass classes to constant temperatures. A) 15oC, B) 20 oC, C) 25 oC, D) 30 oC, and E) 35 oC. Mass classes are represented by solid (light), short-dashed (intermediate), and dotted (heavy) lines. Black circles represent censored observations.

Figure 2-4. Rudbeckia mollis inflorescences during development. A) Aerial view of lower floret anthesis, B) side view of lower floret anthesis, C) distal floret anthesis, D) week 4, E) week 5, F) week 6, G) week 7, H) week 8, I) week 9, and J) week 10 following anthesis. Photographs are courtesy of Nicholas Gerard Genna.

Figure 2-5. Rudbeckia mollis seeds sampled following anthesis. A) Fresh mass, B) dry mass, and C) water content. Data are presented on a single seed basis. Means are compared within developmental weeks with the Bonferroni method at α = 0.05.

Figure 2-6. Rendering of Rudbeckia mollis. Primary branches are determinate and terminal inflorescences are larger than lateral inflorescences. Inflorescences are indeterminate. Leaves are omitted to accentuate branching architecture.

Figure 2-7. Light and transmission electron microscope images of Rudbeckia mollis seeds. Pericarp tissue was removed to aid resin infiltration. A) and D) show the radicle. B) and E) show the apical meristem. C) and F) show the cotyledon region. Abbreviations represent the single cell layer endosperm (En), protein bodies (P), nucleus (N), lipid bodies (L), and starch granules (S). Photographs are courtesy of Nicholas Gerard Genna.

CHAPTER 3 DOES INTRA-POPULATION SEED SURVIVAL DURING BURIAL FOLLOW INTERSPECIFIC RELATIONSHIPS BETWEEN SEED MASS AND COAT THICKNESS?

Introduction

Seeds have three fates in nature: germinate, die, or persist. Germination may occur immediately following shedding in non-dormant seeds or following a season that alleviates inherent dormancy. These species form transient seed banks and are not found in soil samples after one or two years (Walck et al., 2005). Seeds may also die immediately following shedding due to predation or some other stochastic event.

Persistence, on the other hand, involves seeds surviving and not germinating for two or more germination seasons due to dormancy, quiescence, or burial within the soil seed bank to a depth that precludes germination (Thompson et al., 2003). This paper deals specifically with seed longevity in soil, a component of persistence, in an attempt to understand relationships between intraspecific seed mass variation and survival in the soil seed bank.

Seed survival in soil is moderated by pathogen pressure, predation, and inherent longevity (Long et al., 2015). In the short term, inherent longevity may play a small role during in situ burial as transient hydration may allow seeds to repair aging associated damage (Long et al., 2011; Wojtyla et al., 2016). Furthermore, predation has a marked but transient influence on seed survival that decreases following burial (Hulme, 1998).

Pathogen pressure may be the principal factor determining the rate of viability loss in soil and seeds have evolved a variety of physical and chemical defenses to survive. For example, Dalling et al. (2011) propose several seed defense syndromes to characterize broad interspecific relationships between dormancy, defense strategies, and

environmental characteristics. Physical defenses are often attributed to the seed coat and recent research suggests that thicker seed coats can confer increased resistance to pathogen pressure in soil. This relationship was demonstrated at the interspecific (Davis et al., 2008; Davis et al., 2016; Gardarin et al., 2010) and intraspecific (Schutte et al.,

2014; Tricault et al., 2017) levels. All authors found mortality in soil decreased with increasing seed coat thickness. Further, short term survival depends to a greater extent on chemical defenses while long term survival is dependent on physical protection

(Davis et al., 2008; Davis et al., 2016). Therefore, seeds within a population with the thickest seed coats, independent of mass, may have the highest probability of surviving for the greatest duration in soil. However, many questions remain concerning microorganisms and seed bank ecology and attempting to reduce the process to a few variables may be overly simplistic (Chee-Sanford, et al., 2006; Wagner and Mitschunas,

2008).

In the long term, inherent longevity may play a larger role in regulating survival in soil. From an interspecific perspective, species with seeds of higher mass display greater survivorship during burial in Australia (Moles et al., 2003) and the Southeastern

US (Kaeser and Kirkman, 2012). No relationship between seed mass and longevity was found in Alaska (Conn et al., 2006), Austria (Schwienbacher et al., 2010), or France

(Gardarin et al., 2010). Considerable intraspecific variation in longevity also exists during burial. For example, higher mass Coreopsis lanceolata seeds showed higher viability after two years of burial compared to lower mass seeds (Banovetz and

Scheiner, 1984). Similarly, the proportion of viable Galinsoga parviflora ray achenes, with presumed greater mass, exhibited greater viability than disk achenes after two

years of burial (Espinosa-Garcia et al., 2003). Higher mass fruit segments of Raphanus raphanistrum also survived longer during four years of burial (Tricault et al., 2017).

Additionally, recent research indicates that lower mass Rudbeckia mollis (Asteraceae) and Aegilops (Poaceae) seeds lose viability slower than higher mass seeds when subjected to high temperature aging stress (Genna and Pérez, 2016; Guzzon et al.,

2018). Therefore, the possibility exists that intra-population seed mass variation can explain some variability in seed survival during burial and that empirical studies are needed to address discrepancies with high temperature aging, despite research suggesting that high temperature aging may predict seed longevity in soil (Long et al.,

2008). Further, understanding relationships between seed coat thickness and seed mass can clarify factors connected to survivorship in soil, as discussed by Schutte et al.

(2014).

I chose Rudbeckia mollis L. (Asteraceae) for this research due to its annual life history and lack of physical dormancy. I buried seeds of different mass to a depth that would largely inhibit germination and proposed three hypotheses: 1) seeds of different mass will display differing short term survivorship in soil, 2) survival in soil will increase with pericarp thickness, and 3) survival in soil will follow a negative sigmoid distribution.

The last hypothesis addresses a knowledge gap in seed bank ecology. Persistence in soil is described by a negative exponential relationship (Murdoch and Ellis, 2000).

However, persistence in this case includes seeds leaving the soil seed bank due to germination or predation. For seeds buried deeply in soil to minimize germination, it is unclear if viability loss occurs in a negative sigmoid fashion in agreement with viability loss patterns in ex situ storage.

Materials and Methods

Plant Material and Establishing Mass Classes

I used all seed material and mass classes established in Chapter 2 for all research described here except for allocation to physical defenses. I obtained pericarp thickness data from Rudbeckia mollis seeds collected in the fall of 2013 from a wildflower producer (Wildflowers of Florida Inc., Alachua, FL). Mean seed mass was

393 ± 35, 423 ± 29, and 474 ± 38 µg in the light, intermediate, and heavy classes, respectfully. In this seed lot, mass was not significantly different in the light and intermediate classes whereas both were significantly lower in mass compared to the heavy class (F2,42 = 24.2, P <0.0001).

Seed Burial

All seeds were counted by mass by multiplying mean seed mass in each class by the number of seeds desired. A total of approximately 4500 seeds from each mass class were separated into 15 samples of about 300 seeds each. I mixed all samples with 100 grams of autoclaved sand (van Mourik et al., 2005) and incorporated the mixture within two 7.5 X 10 cm polyester organza bags. I tagged and assigned all bags in a completely randomized fashion within a 2 X 2 m2 area within a larger non-irrigated wildflower plot on the University of Florida campus. All bags were buried 30 cm apart and to a depth of 25 cm on April 7, 2017. I recovered one bag from each mass class every 28 days for 168 days then every 62 days for 558 days. This roughly translates to retrieval every 1-6, 8, 10, 12, 14, 16, 18, 20, 22, and 24 months. Data are presented as months.

A data logger recoded air temperature, relative humidity, soil temperature, and soil moisture in the burial plot. I buried the temperature and soil moisture probes at the same depth as burial bags. I present data from January 1-December 31, 2018.

Germination Testing Following Burial

I recovered burial bags and allowed the sand-seed mixture to dry on a laboratory benchtop for one day before sieving seeds with wire mesh. Next, I counted filled and empty seeds after first determining fill with a press test. Seeds were discarded after 1-3 months of burial and a filled and empty seed count was not obtained. I randomly selected 100 filled seeds from each mass class for a viability assessment. Samples were separated into four, 25 seed sub-samples then submitted to a germination test inside an incubator set to 25oC with a 12 hour photoperiod for 28 days. Seeds were sown on top of blue blotter paper hydrated with autoclaved deionized water inside plastic boxes. I recorded germination daily and water was added to germination boxes as necessary. I selected 25oC as previous research showed this was the optimal constant temperature for R. mollis germination (Chapter 1). I filled sub-samples sequentially in the few instances that burial bags had fewer than 100 filled seeds; meaning some subsamples had fewer than 25 seeds. I determined germination lag

th th th time, rate (1/t50), uniformity (t75-t25), 25 , 50 , and 75 percentiles from daily germination counts.

Allocation to Physical Defenses

Rudbeckia mollis achenes are square when examined transversely (Genna,

2015). Fifty seeds from each mass class were embedded in cyanoacrylate and supported by a cork stub and allowed to cure for 24 hours. I mounted transverse free hand sections in water and examined sections in a compound microscope at 400X. I

measured pericarp thickness on all four sides, towards the middle, and at a point where the repeating convex semi-circular shaped pericarp segments were widest (Genna,

2015). I averaged all measurements for each seed and means are presented.

Statistical Analyses

I analyzed pericarp thickness data with linear regression. Assumptions of normality and constant variance were determined graphically. I analyzed filled, empty, and total seed count data with categorical data analysis according to Stokes et al.

(2012). Mass classes were coded as 1, 2, and 3 corresponding to the light, intermediate, and heavy classes, respectfully. I generated the general association statistic to test the null hypothesis of no association between counts at each burial month and mass class. I generated the nonzero correlation statistic to test the null hypothesis of a linear relationship between counts at each burial month and mass class.

Finally, I determined correlation between months of burial and mass class with

Spearman rank correlation. I also correlated soil temperature against final germination percentage and germination rate for all mass classes with Pearson correlation.

I analyzed germination data following burial with non- and semi-parametric time to event analyses. Local extinction events (defined as months with less than 100 filled seeds recovered in a given bag) occurred in months 10, 16, 20, and 22 for the light class, months 18 and 22 for the intermediate class, and month 20 for heavy class seeds. The number of filled seeds was less than 100 for each of these data points and germination percent was abnormally low. Therefore, these data were excluded from analyses. Germination was the event of interest and coded as 1. Moldy seeds removed before 28 days and non-germinating seeds after 28 days were censored and coded as

0. I built a Cox regression model with mass as a class variable and month of burial as a

continuous variable. I verified the proportional hazards assumption with graphical and residual analyses. I used orthogonal contrasts to compare mass classes within burial months.

Results

Air and Soil Climate

Air and soil temperature followed a similar pattern in 2018. However, air temperature fluctuated considerably more with an average weekly range of 22oC throughout the year. Air temperature reached a minimum of -7.2oC during week 1 and peaked at 41.5oC during week 27. In contrast to air temperature, soil temperature at 25 cm below the surface remained relatively constant with a mean weekly range of 3oC.

Soil temperature was lowest during week 1 at 9.3oC and reached a maximum of 32.5oC during week 32. Maximum air relative humidity approached 100% during the entire year with an average of 73 ± 10%. Soil water potential fluctuated between seasons and consisted of two dry periods and two wet periods. Soils were wettest from weeks 0-8,

21-37, and 46-52 and fluctuated from 0 to -69 kPa corresponding to summer and winter seasons. Soils were driest from weeks 8-21 and 37-46 and fluctuated from 0 to -200 kPa during spring and fall (Figure 3-1).

Seed Count Data

The number of filled seeds recovered from burial bags decreased and the number of empty seeds increased with time. From months 4-8, the number of filled seeds ranged from 176-229 in all mass classes. Similarly, the number of empty seeds ranged from 62-111 during this period. From months 10-24, stochastic local extinction events occurred in all classes that significantly reduced the number of filled seeds and increased the empty seed count. These events occurred at months 10, 16, 20, and 22

for the light class, months 18 and 22 for the intermediate class, and month 20 for heavy class seeds (Figure 3-2 A-B). There was a significant general association between month of burial and mass class for filled and empty seed counts. Filled seed counts also displayed a significant nonzero correlation while empty seed counts did not. The general association and nonzero correlation statistics were not significant for total seed count, indicating that the approximate 300 seeds incorporated into bags were recovered each month (Figure 3-2C, Table 3-1). Spearman rank correlation for filled seed counts showed a weak positive relationship between mass class and burial months, indicating that higher mass seeds were more likely to be recovered filled. Conversely, a weak negative correlation described the empty seed count indicating that lower mass seeds were more likely to be recovered empty, but this was not significant. There was no significant correlation between total seed count and months buried (Table 3-2).

Germination Following Burial

Local extinction events discussed prior become evident in final germination data

(Figure 3-3A). Final germination percentage ranged from an average minimum and maximum of 28-98% with an overall mean of 83% across all mass classes and burial months when local extinction events were excluded (Figure 3-3B). Final germination percentage increased from months 1-8 and decreased to a minimum at month 10. Final germination percentage increased again to a maximum at month 14 and decreased steadily again from months 14-24. Cox regression modeling showed that both mass class and burial month significantly influenced germination. Overall, the probability of germination decreased by 2.0% for each additional month of burial across classes

(Table 3-3). Mass class comparisons within burial months reveled that most differences occurred early in the burial experiment. Intermediate seeds had the highest probability

of germination at month 1, light seeds had the highest probability of germination from months 2-4, and heavy class seeds had the highest probability of germination at months

5-6 and 8. No significant differences between mass classes were detected between months 12-22. Germination diverged between mass classes at month 24 with light class seeds displaying the highest and heavy class seeds displaying the lowest probability of germination (Table 3-4).

The underlying pattern in germination response was confirmed in germination summary statistics over the 24-month burial period. The undulating pattern in germination rate and final germination percentage was the inverse of lag time, uniformity, and percentile data. For example, as germination rate tended to increase from months 1-5 and 10-16, uniformity tended to improve. Conversely, germination was less uniform when germination rate tended to decrease from 5-10 and 16-22 months.

No clear difference in any germination parameter is visually identifiable when mass classes are compared (Figure 3-4). However, germination rate and percentage for all mass classes positively correlated with soil temperature over the 24-month burial period

(Table 3-5).

Allocation to Physical Defenses

Pericarp thickness ranged from 24.7-41.4 µm with a mean (± SE) thickness of

32.2 ± 3.0 in all seeds sampled. There was a weak positive relationship between

2 pericarp thickness and mass in R. mollis seeds (F1,35 = 35.43, P <0.0001, R = 0.19).

The relationship between pericarp thickness and seed mass is described by the linear relationship: (pericarp thickness (µm)) = 24.59 + 0.018 (seed mass(µg)) (Figure 3-5).

Discussion

This research demonstrated that R. mollis seeds are capable of surviving for at least two years in soil and that short-term survival (defined as mass-dependent viability loss) is not dependent on seed mass or coat thickness, rejecting hypotheses one and two of this research. However, there does appear to be a weak but significant relationship between seed mass and extinction events documented in this research.

Light class seeds experienced four local extinction events, intermediate seeds experienced two, and heavy class seeds experienced one. This may be a coincidence, but evidence for a coincidence is low since burial bags that experienced extinction events were not co-located in the randomized burial plot. Furthermore, evidence for a relationship between pericarp thickness and local extinction events is high since pericarp thickness increases with seed mass in R. mollis. Therefore, higher mass seeds with thicker pericarps may be less susceptible to pathogen pressure in soil due to greater physical defenses.

Seed defense syndromes introduced by Dalling et al. (2011) suggest that seeds with physiological dormancy may rely on both chemical and physical defenses to survive in a high pathogen pressure environment characterized by higher moisture availability. Research presented here supports Dalling et al. (2011) on an intra- population basis regarding specifically physical defenses since chemical defenses in R. mollis were not assayed. However, 16 tree species in Panama showed a clear relationship between permeability and defense strategies suggesting that species with permeable seed coats may rely on chemical defenses more than physical defenses

(Zalamea et al., 2018). This research highlights the necessity to apply established interspecific relationships between seed mass, coat thickness, chemical defenses, and

survival in soil to intraspecific or intra-population studies to either support or refute broad interspecific conclusions.

Interestingly, seeds buried in soil in this study have greater longevity than R. mollis seeds stored in a non-climate controlled shed in Florida in a concurrent study

(Chapter 4). This demonstrates that seeds can potentially repair damage accrued over time from transient rehydration (Long et al., 2015). For instance, moving seeds from a high temperature aging treatment to a priming treatment restored antioxidant capacity and improved germination in Avena sterilis (Long et al., 2011). Long et al. (2011) also documented high glutathione capacity in field buried seeds, although this was not correlated with survival in soil due to the interaction with soil microorganisms. Transient hydration in the form of priming may also allow seeds to repair DNA damaged following desiccation (Wojtyla et al., 2016). In this case, R. mollis seeds appeared to take advantage of wetting and drying cycles to maximize longevity in soil. Research suggests that repair will continue until all resources are exhausted (Kranner et al.,

2010); although it is unclear why seeds may reach exhaustion during burial or how exhaustion is defined.

I recovered an average of 290 seeds across classes and sampling periods over the 24-month burial study. This number approached the approximate 300 seeds per bag included before burial. It is not possible to know how many seeds were lost during the burial experiment, but the proportion was probably small. The average number of filled seeds recovered each sampling month was 186 and the average number of empty seeds was 104 across classes excluding outlying data points. This means that approximately 100 seeds fatally germinated or were consumed by soil pathogens in

each bag during the two-year burial period. I speculate that a majority of the 100 empty seeds were a result of fatal germination since the empty seed count was consistent.

Seeds in this research were buried eight months following harvest and were no longer conditionally dormant (Chapter 1). Therefore, a flush of germination events inside burial bags following burial may have occurred. Fatal germination is a challenge for seed burial studies as burial depth and fatal germination are negatively correlated (Long et al., 2015). Burying R. mollis deeper and initiating the burial study immediately following seed collection when inherent dormancy is highest may have reduced the empty seed count.

Addressing my third hypothesis, that seed viability loss will occur in a negative sigmoid fashion, conflicting viability loss patterns are reported in the literature. For example, 30 species across three studies demonstrated a negative exponential decline in viability for seeds buried in soil from 27 months to 20 years (Conn et al., 2006;

Espinoza-Garcia et al., 2003; Kaeser and Kirkman, 2012). Further, Wagner and

Mitschunas (2008) assumed a negative exponential viability decline when modeling seed deaths in soil. To the contrary, 17 weedy species lost viability in a negative sigmoid fashion over two to four years (Gardarin et al., 2010; Masin et al., 2006; Tricault et al., 2017). In agreement with van Mourik et al. (2005), I suspect that the shape of the seed mortality curve depends on the mode of death in soil. For example, germination phenology studies that permit germination from soil due to shallow burial typically demonstrate a negative exponential decline in persistence (Davis et al., 2016).

Similarly, in studies utilizing burial bags, the survivorship curve may also be negative exponential if seeds are succumbing to a combination of pathogen pressure and

physiological aging. On the other hand, if physiological aging is the sole or primary cause of seed mortality in soil, then I suspect the survivorship curve may resemble a typical negative sigmoid longevity curve as observed for seed viability loss during ex situ germplasm storage. This can be tested by isolating variables in a burial study, perhaps by performing a simultaneous study comparing seeds in bags buried in field conditions to seeds buried in sterile sand within growth chambers.

Despite the lack of association between mass and longevity, R. mollis is capable of sensing seasonality and the undulating pattern of germination percentage and rate correlated positively with soil temperature. Seeds with non-deep physiological dormancy typically exhibit dormancy cycling during burial in a phenomenon called the dormancy continuum (Baskin and Baskin, 2014). In the case of R. mollis, a summer annual that germinates in spring and completes its life cycle in fall, seeds are shed conditionally dormant and exhibit a mass-dependent germination response. When exposed to colder fall (27/15oC) and winter (22/11oC) temperatures, heavy class seeds germinate to approximately 70% while light class are capable of germinating to approximately 90%

(Chapter 1). Rudbeckia mollis becomes nondormant after a period of dry after-ripening when seeds are stored under room temperature conditions (Genna and Pérez, 2016).

However, when buried in soil, conditional dormancy is re-imposed during the summer and alleviated during the winter. There does appear to be a mass-dependent germination response early following burial from months 1-4 based on final germination percentages. However, seeds in this research were buried in April while wild R. mollis seeds are typically shed in September. Therefore, it is not possible to draw any broad conclusions about the reinstatement of mass-dependent conditional dormancy since I

buried seeds eight months following shedding. However, this does merit repeating, since to my knowledge, I do not believe there has been a documented case of mass- dependent dormancy cycling reported in the literature.

Conclusion

This research provides tentative support for the “physiological dormancy defense syndrome” proposed by Dalling et al. (2011) since R. mollis longevity in soil may be related to seed coat thickness and possible unknown chemical defenses. More intraspecific (intra-population or intra-accession) studies are needed to address the relationship between seed defenses and survivorship in soil within wild plant populations. Specifically, research addressing seed mass, coat thickness, and chemical defenses are needed to either support or refute broad interspecific claims. More broadly, intraspecific research can shed light on weed seed dynamics in agricultural plots and on the seed ecology of native species in natural seed banks. For instance, smaller seeded species are more likely to be incorporated in the soil seed bank

(Thompson et al., 1993). Within a single species however, it is not currently known if a subpopulation of seeds of a particular mass are the primary contributors to long term persistence in natural soil seed banks.

Table 3-1. Nonzero and general association statistics for filled, empty, and total seed counts of Rudbeckia mollis seeds buried for 24 months. Count Statistic df Value P Filled Nonzero correlation 1 14.11 0.0002 General association 22 823.0 <0.0001 Empty Nonzero correlation 1 2.539 0.1110 General association 22 419.1 <0.0001 Total Nonzero correlation 1 0.0102 0.9198 General association 22 3.240 1.000

Table 3-2. Spearman correlation statistics for filled, empty, and total seed counts correlated with Rudbeckia mollis mass classes buried for 24 months. Count Correlation (r) SE P Filled 0.0586 0.0134 <0.0001 Empty -0.0156 0.0145 0.2813 Total -0.0012 0.0098 0.8991

Table 3-3. Cox regression model for the germination response of Rudbeckia mollis seeds of different mass following 24 months of burial. Covariate (xi) Coefficient (βi) SE of βi Wald 2 P Hazard Ratio Mass class -a - 19.16 <0.0001 - Month -0.020 0.0023 73.79 <0.0001 0.980 aMain effect coefficients for class variables are not calculable.

Table 3-4. Selected orthogonal contrast comparisons of Rudbeckia mollis mass classes within burial months. Burial months 12-22 are excluded as contrasts were not significant. Confidence limits that exclude one are significant at α = 0.05. Month Comparison Hazard Ratio 95% Confidence limits 0 Light vs Intermediate 1.019 (0.755, 1.377) Light vs Heavy 1.156 (0.856, 1.561) Intermediate vs Heavy 1.134 (0.841, 1.529) 1 Light vs Intermediate 0.948 (0.698, 1.289) Light vs Heavy 1.363 (0.990, 1.877) Intermediate vs Heavy 1.437 (1.048, 1.972) 2 Light vs Intermediate 1.251 (0.921, 1.701) Light vs Heavy 1.566 (1.148, 2.137) Intermediate vs Heavy 1.252 (0.914, 1.714) 3 Light vs Intermediate 1.629 (1.203, 2.206) Light vs Heavy 1.819 (1.344, 2.463) Intermediate vs Heavy 1.116 (0.820, 1.521) 4 Light vs Intermediate 1.338 (0.995, 1.799) Light vs Heavy 1.541 (1.138, 2.087) Intermediate vs Heavy 1.152 (0.850, 1.562) 5 Light vs Intermediate 0.537 (0.400, 0.721) Light vs Heavy 0.521 (0.389, 0.699) Intermediate vs Heavy 0.970 (0.721, 1.306) 6 Light vs Intermediate 0.331 (0.242, 0.453) Light vs Heavy 0.408 (0.298, 0.558) Intermediate vs Heavy 1.231 (0.922, 1.645) 8 Light vs Intermediate 0.573 (0.427, 0.770) Light vs Heavy 0.332 (0.246, 0.448) Intermediate vs Heavy 0.579 (0.430, 0.780) 10 Light vs Intermediate -a - Light vs Heavy - - Intermediate vs Heavy 1.579 (1.136, 2.194) 24 Light vs Intermediate 1.432 (1.016, 2.017) Light vs Heavy 4.267 (2.753, 6.615) Intermediate vs Heavy 2.980 (1.904, 4.664) aLight class germination was an outlier at 10 months and removed from analyses.

Table 3-5. Pearson correlation statistics for Rudbeckia mollis final germination percentage and germination rate correlated with soil temperature following burial for 24 months. Variable Correlation (r) P Germination percentage 0.3707 0.0170 Germination rate 0.3352 0.0345

Figure 3-1. Air and soil conditions in the burial plot measured from January 1 to December 31, 2018. A) Air temperature, B) soil temperature, C) air relative humidity, and D) soil water potential. Soil temperature and soil water potential were measured 25 cm below the soil surface at the same depth as burial bags. Solid lines represent mean values. Dashed lines represent minimum and maximum recorded values. The months buried axis indicates the month of burial that corresponds to air and soil conditions during 2018.

Figure 3-2. Counts of Rudbeckia mollis seeds from the light, intermediate, and heavy mass classes following 24 months of burial. A) Filled, B) empty, and C) total seed counts. Empty seeds were determined with a press test. Total seed count = filled + empty.

Figure 3-3. Final germination percent data for Rudbeckia mollis mass classes following 24 months of burial. A) With outlying local extinction events and B) without outlying local extinction events.

Figure 3-4. Germination summary statistics for Rudbeckia mollis mass classes following 24 months of burial. A) Lag time, B) 25th percentile, C) germination th th rate (1/t50), D) 50 percentile, E) germination uniformity (t75 - t25), and F) 75 percentile.

Figure 3-5. Relationship between pericarp thickness and seed mass of Rudbeckia mollis seeds separated into a light, intermediate, and heavy mass class. Linear regression fitted the equation for all data points: (pericarp thickness (µm)) = 24.59 + 0.018 (seed mass(µg)).

CHAPTER 4 NO EVIDENCE OF MASS DEPENDENT SEED LONGEVITY FOLLOWING STORAGE OR REPEATED RELATIVE LONGEVITY ASSESSMENTS

Introduction

Ex situ seed conservation is critical for wild plant species with agronomic and ecological importance (Castañeda-Álvarez et al., 2016; Mounce et al., 2017). Unlike domesticated species, seed lots from wild species are inherently heterogeneous with marked phenotypic and genotypic variation that complicates successful seed storage

(Hay and Probert, 2013; Walters, 2015). Seed mass is one phenotypic trait that varies widely within seed lots and has garnered much attention in the ecological literature

(Chapter 2, 3). In the context of seed conservation, empirical research established no relationship between interspecific seed mass variation and seed storage longevity but documented marked intra-accession longevity variation (Walters et al., 2005). This presents an opportunity to examine mass-specific longevity variation within wild plant accessions that is masked by research focused at the interspecific level.

A current problem for seed conservation is understanding why longevity varies across species and is relatively conserved within species. For instance, the time necessary to reduce germination to 50% (P50) ranged from 7-633 years across 276 species stored under genebank conditions (4-8% moisture content, 5 or -18°C), while

P50 values for different species assayed across different studies were comparable

(Walters et al., 2005). This implies that species-specific genetics and downstream seed chemistry drive differential longevity. The same concept is potentially relevant to variable seed longevity within accessions, since physical and physiological characteristics can vary widely within wild plants. In the context of seed storage, intra- accession longevity may be moderated in part due to mother plant effects driven by

developmental location. For example, seeds located closer to resources display greater mass and higher mass seeds often contain higher proportions of macro- and micronutrients compared to lower mass seeds (Obeso, 2012; Vaughton and Ramsey,

1997). However, no interspecific link has been established between inherent macro- and micronutrients and longevity in storage (Walters et al., 2005).

Differences in seed maturity may also drive intra-accession longevity variation in wild seed lots as longevity is acquired during maturation drying and after seeds have become desiccation tolerant (Hay and Probert, 2011; Leprince et al., 2017). For example, P50 values generated following controlled deterioration assays were significantly higher for mature seeds from two wild species compared to immature seeds (Probert et al., 2007). This contrasts the historical assumption that seed quality is highest at mass maturity and declines thereafter (Ellis, 2019). Hay and Probert (2011) recommend collecting wild seeds as close to natural dehiscence to avoid maturity related longevity affects. However, optimizing wild seed collection is not always possible because the timing of maturity can vary between years and high collection costs prohibit routine population monitoring.

Long-term seed storage in the cold (e.g. ≤ -18°C) is possible for most plant species due to desiccation tolerance acquired during development. Intracellular compounds including late embryogenesis abundant proteins, heat shock proteins, and non-reducing sugars accumulate during maturation drying and contribute to intracellular

-1 glass formation at water contents below 0.10 g H20 · g dry mass (Leprince et al.,

2017). Intracellular glasses act as stabilizing forces for proteins and lipid membranes during desiccation but do not confer desiccation tolerance as desiccation intolerant

seeds also form glasses (Buitink and Leprince, 2004; Buitink and Leprince, 2008;

Hoekstra et al., 2001).

Bulk storage reserves are also involved in desiccation tolerance by minimizing cell membrane perturbations following water loss (Walters et al., 2002; Walters and

Koster, 2007). Likewise, specific types and amounts of reserves such as oil content drive equilibration water content (Probert, 2003; Vertucci and Roos, 1990). Small differences in oil content and correspondingly small differences in equilibrium moisture content within wild plant accessions may translate into quantifiable longevity differences for seeds of different mass. For example, percent oil content and seed mass decrease acropetally within sunflower inflorescences (Hassan et al., 2011). In contrast, oil content increases and seed mass decreases with greater distance from the main stem in

Lupinus albus (Crochemore et al., 1994). In another example, Dwivedi et al. (1990) showed a positive relationship between oil content and seed mass in 33 peanut genotypes. These examples highlight how seed mass and oil content may correlate, although in different directions depending on the species, and may drive differing equilibrium moisture content during storage.

Seed longevity is moderated by temperature and water content and is extended as both are reduced (Ellis and Roberts, 1980). Mechanisms underpinning seed deterioration depend on seed moisture status. Seeds stored in a high relative humidity

(RH) and high temperature environment corresponding to hydration levels III-V lose viability rapidly due to reactive oxygen species generated by metabolic machinery

(Walters, 2005). Reactive oxygen species attack and render membrane lipids, DNA, and proteins nonfunctional, but mechanisms are not conserved across species and

change with seed moisture content (Corbineau et al., 2002; El-Maarouf-Bouteau et al.,

2011; Lehner et al., 2008). To the contrary, seeds stored cold and dry are metabolically inactive and lose viability in unresolved deteriorative processes regulated by molecular mobility. Intracellular glass formation is believed to slow deterioration by greatly reducing molecular mobility and the rate of deteriorative reactions (Buitink and Leprince,

2004; Buitink and Leprince, 2008; Walters, 2005). However, molecular mobility is not zero in extreme cold and seeds ultimately lose viability after extended storage as glasses relax and deteriorative reactions occur (Ballesteros et al., 2018; Walters, 1998;

Walters et al., 2004; Walters et al., 2010).

Predicting longevity in storage is the ultimate challenge in seed conservation.

Seeds stored in cold and dry conditions may maintain viability for decades or centuries making empirical seed storage research difficult and expensive (Walters et al., 2005).

High temperature and high humidity conditions, so called artificial aging or accelerated aging, are used to gain insight into seed deterioration during cold and dry storage despite mechanisms driving viability loss being different in both conditions (Ballesteros and Walters, 2011; Schwember and Bradford, 2010). Controlled deterioration is widely adopted today as a comparative longevity standard by rapidly aging seeds at 45oC in a

60% RH environment (Newton et al., 2009). Controlled deterioration may predict broad scale interspecific and intraspecific longevity on a comparative basis and may highlight populations that are relatively long or short lived. For example, Probert et al. (2009) demonstrated a significant correlation of longevity rankings between seeds stored in genebank conditions and seeds subjected to controlled deterioration. Merritt et al.

(2014) and Probert et al. (2009) also demonstrated that non-endospermic or serotinous

seeds from hot environments have the greatest longevities compared to endospermic seeds originating from wetter climates. Kochanek et al. (2009) and Schoeman et al.

(2010) demonstrated that P50 values generated with controlled deterioration can vary widely among populations within species. However, it is unclear if comparative longevity assays also function on narrower scales, such as predicting relative longevity within accessions for seeds of different mass.

Recent research showed lower mass seeds within populations in the genus

Aegilops and Rudbeckia had greater longevity than higher mass seeds following high temperature aging (Genna and Pérez, 2016; Guzzon et al., 2018). Satyanti et al. (2018) also showed a negative interspecific relationship between longevity and seed mass in

56 Australian species using controlled deterioration. This research contrasts Probert et al. (2009) that found no relationship in 195 species globally, Merritt et al. (2014) that found a weak positive relationship in 172 species in Australia, and Mira et al. (2019) that found greater longevity in higher mass seeds within a population of Hirschfeldia.

Empirical storage research is therefore needed to test claims purported by high temperature aging.

In this research, I attempt to elucidate relationships between seed mass and longevity within a single accession. I accomplished this by separating Rudbeckia mollis

(Asteraceae) L. seeds into distinct mass classes followed by storage under four different conditions of varied temperature and RH stress for two years. I also subjected different mass seeds to repeated saturated salt accelerated aging (SSAA) treatments over time to understand if fresh seed aging is predictive of viability loss during storage and is consistent with SSAA treatments as seeds age. I chose R. mollis as I have previously

characterized its biology (Chapter 2, 3; Genna and Pérez, 2016). I hypothesize that: 1) seeds of different mass will deteriorate at different rates in short term storage, 2) comparative longevity following SSAA is predictive of mass-dependent deterioration in storage, and 3) comparative longevity following SSAA assays will not change with increasing storage duration.

Materials and Methods

Plant Material and Establishing Mass Classes

I used all seed material and mass classes established in Chapter 2 for all research described here. Refer to Chapter 2 for more details.

Water Sorption Isotherms

I generated water sorption isotherms with fresh seeds from each mass class on

October 4, 2016. A total of 2400 randomly selected seeds from each mass class were divided into twelve samples of 200 seeds each and randomly assigned to one of twelve saturated salt solutions including KNO3, KCl, NaCl, Ca(NO3)2, NaI, MgCl2, CaCl2, KAc,

LiCl, KOH, ZnCl2, and P2O5. Salts were contained within sealed glass desiccators and

o kept at room temperature (23 C). Except for KOH and P2O5, I prepared each saturated salt solution by mixing distilled deionized water with each salt to create a slurry. I prepared KOH by adding distilled deionized water to KOH pellets until fusion occurred.

No water was added to P2O5 pellets. The 200 seed samples were separated into eight subsamples of 25 seeds within each desiccation jar. I measured fresh mass daily on one sub-sample from each class within each desiccation jar. After three days of consistent fresh mass measurements, I measured fresh mass on four sub-samples from each class and placed each subsample in a 100oC oven for three days to determine dry mass. I determined water content on a dry mass basis. I used the remaining four sub-

100

samples to evaluate germination following drying by placing seeds inside plastic germination boxes on blue blotter paper hydrated with autoclaved (117.7 kPa, 121 °C,

40 minutes) deionized water. Germination boxes were placed inside a 25oC incubator with a 12 hour photoperiod for 28 days. I recorded germination daily and water was added to germination boxes as necessary.

Seed Storage Treatments

I counted all seeds on a mass basis by multiplying mean seed mass in each class by the number of seeds desired. Approximately 8,000 seeds from each mass class were separated into four groups of about 2,000 seeds. The groups were assigned to one of four storage treatments including a non-climate controlled outdoor shed

(variable temperature and RH), climate-controlled room (23oC, 42.5% RH), refrigerator

(4oC, 15% RH), or freezer (-18oC, 15% RH). Each seed group was separated into 16 samples of 125 seeds each. I placed each sample assigned to the room and shed treatments inside glass vials with threaded plastic caps. I chose glass vials to allow seed moisture to fluctuate since the plastic caps did not have rubber gaskets and were not presumed to be airtight. Seeds were sealed inside foil laminate bags (Moore and

Buckle Flexible Packaging, Saint Helens, UK) for the 4oC and -18oC temperature treatments after seven days of water content adjustment with saturated LiCl (13.0% RH,

23.5oC). Seed water content after equilibration averaged 3.5 ± 0.14% across classes.

Seeds were placed in all storage treatments on April 7, 2017. I retrieved one bag or vial from each mass class and storage treatment every 28 days for 168 days and every 62 days for 558 days. This roughly translates to retrieval every 1-6, 8, 10, 12, 14,

16, 18, 20, 22, and 24 months. Data is presented as months. I evaluated viability by randomly selecting 100 seeds from each sample and dividing the sample into four sub-

101

samples of 25 seeds for germination inside a 25oC incubator with a 12 hour photoperiod for 28 days. I recorded germination daily and water was added to germination boxes as necessary. I selected 25oC as previous research showed this was the optimal constant temperature for R. mollis germination (Chapter 1).

A data logger recorded air temperature and RH inside the outdoor shed. I present data from January 1-December 31, 2018.

Saturated Salt Accelerated Aging (SSAA)

I used SSAA as a comparative estimate of seed longevity over the two-year storage period. I first subjected seeds from each mass class to SSAA on September 15,

2016 when seeds were fresh. I also performed SSAA on April 7, 2017, or month 0 of the storage experiment to assess vigor loss during the eight months before beginning the storage experiment. Seeds stored in the room and inside the shed were also subjected to SSAA at 6, 12, 18, and 24 month intervals. Refrigerator and freezer stored seeds were not subjected to SSAA due to insufficient seed.

For each combination of storage duration and storage location, 1,250 seeds from each mass class were separated into 10 samples of 125 seeds. I suspended each 125 seed sample over 60 mL of saturated NaCl with a mesh screen inside a plastic box

(Hoffman Manufacturing Inc., OR, USA). I created saturated NaCl by heating deionized water to 45oC followed by the incremental addition of NaCl salt until saturation. Seeds were removed from a 41.0oC incubator after 0, 5, 10, 15, 20, 25, 30, 35, 40, and 45 days. After each SSAA duration, 100 randomly selected seeds from each mass class were divided into four sub-samples of 25 seeds for germination inside a 25oC incubator with a 12 hour photoperiod for 28 days. I recorded germination daily and water was added to germination boxes as necessary.

102

Statistical Analyses

I estimated the relationship between water content and RH in each mass class with nonlinear regression specifying a cubic polynomial function in SigmaPlot (Systat software, Inc., San Jose, CA, USA). I analyzed all germination data with probit analysis and estimated the time for viability to fall to 50% by fitting the Ellis and Roberts (1980) viability equation:

v = Ki – (p/σ) where v equals the viability of the seed lot after p days in storage in probits, Ki is the initial viability of the seed lot in probits, and σ represents the time in days for viability to fall by 1 probit (i.e. the standard deviation of the normal distribution of seed deaths over time).

Results

Water Sorption Isotherms

All mass classes displayed a sigmoid shaped water sorption isotherm. Seeds of the light, intermediate, and heavy classes had an average initial water content of 6.89 ±

0.15% corresponding to a water potential of about -63.2 MPa (Figure 4-1). Final germination percent was ≥ 85% in all mass classes following equilibration to all RH environments ranging from 0.5 to 91% (Table 4-1).

Seed Storage Treatments

Air temperature in the outdoor shed ranged from an average low of -0.5oC in weeks one and three to 34.2oC during week 37. Relative humidity ranged from an average low of 29.7% in week 10 to 73.8% in week 29 (Figure 4-2).

Shed stored seeds exhibited a marked viability decline over the 24-month period.

Viability was ≥ 84% in all classes from 0-5 months. From 6-14 months, viability was ≥

103

81% in the heavy class while light class seed viability reduced to 57% at 14 months.

Viability in all classes was erratic and independent of mass from 16-24 months ranging from 0-79%. Averaging final germination percentage data for all classes yielded viability equation parameters: Ki = 7.69 probits,  = 5.43 probits, P50 = 14.6 months (Figure 4-

3A). On the contrary, seeds from all mass classes stored in room, 4oC, and -18oC treatments did not exhibit any viability loss during the 24-month storage period. In all cases, final germination percentage ranged from 81-97% (Figure 4-3 B-D).

Saturated Salt Accelerated Aging

All R. mollis mass classes exhibited similar relative longevity following SSAA when fresh and at month 0 (Figure 4-4). However, P50 decreased from an average of

32.2 days across classes when fresh to an average of 28.5 days at month 0 (Table 4-2).

Subjecting room and shed stored seeds to repeated SSAA over two years revealed storage location and mass-dependent comparative longevities. For seeds stored in the shed, mass-dependent longevity was detected at 6 and 12 months (Figure

4-5 A, C). At month 6, light and intermediate class seeds displayed an average P50 of

18.3 days, a value 3.3 times greater than P50 of heavy seeds. In contrast to month 6, intermediate class seeds lost viability most rapidly following 12 months of storage in the shed with a P50 that was 13.0 times lower than light or heavy seeds. SSAA deterioration curves for shed stored seeds were similar across different mass classes by month 18

(Figure 4-5E), albeit with comparatively reduced P50 values compared to previous storage durations. Germination did not exceed 50% in the intermediate and heavy classes at month 24 and P50 was only 1.26 days in the light class (Figure 4-5G, Table 4-

3).

104

Alternatively, relative longevity was mass independent for seeds stored in the climate controlled room over 24 months. P50 was also an average of three times greater at each six-month assessment when averaged across classes for seeds stored in the room compared to the shed. Finally, longevity curves changed from negative sigmoid to negative exponential at month 24 in the room due to the loss of the initial asymptomatic period (Fig. 4-5B, D, F, H; Table 4-3).

Discussion

This research attempted to demonstrate that seed mass would explain some component of intra-accession longevity variation described in long-term storage research (Walters et al., 2005). I posed three hypotheses to address this problem. First, that seeds of different mass would deteriorate at different rates in storage. My results here do not support this hypothesis for seeds stored in a non-climate-controlled shed and my results are inconclusive for colder and drier treatments as seeds did not lose viability. Second, relative longevity assessed with SSAA would predict deterioration in storage. My results here tentatively support that hypothesis for seeds stored in a non- climate-controlled shed since viability loss was mass independent and are inconclusive for colder treatments. Finally, that mass-dependent relative longevity following SSAA would not change over time; meaning, if P50 values were relatively similar among classes in fresh seeds, then they should be relatively similar at each six-month assessment as well. Again, my results tentatively support that hypothesis based on seeds stored in the room and are less conclusive regarding the shed treatment. Below, I will discuss these conclusions in the larger context of this research.

Water sorption isotherms generated with fresh seeds demonstrated that equilibrium moisture content is independent of mass in R. mollis. This implies that light,

105

intermediate, and heavy mass seeds have similar lipid content since oil content drives equilibration water content (Probert, 2003; Vertucci and Roos, 1990). In Chapter 2, I demonstrated that different mass R. mollis seeds have similar lipid proportions when surface area was assessed with ultramicrotomy. However, lipid content as a percentage of fresh mass was not assessed in each mass class and R. mollis lipid content is not reported in the literature on a mass basis. Collectively, this does not suggest that lipid content does not vary within R. mollis populations since lipid content was shown to vary with seed mass in other species (Crochemore et al., 1994; Dwivedi et al., 1990; Hassan et al., 2011). Rather, this suggests that the way the light, intermediate, and heavy mass classes were created may mask differences in lipid content within R. mollis populations.

For example, lipid content may not correlate with mass across a population of seeds but may vary within mother plants as in sunflower and Lupinus albus (Crochemore et al.,

1994; Hassan et al., 2011). Lipid content and variable equilibrium moisture content may still explain some component of intra-accession longevity variation in other species if seeds of known lipid content are followed through storage time. This would require correlating lipid content with a nondestructive tool such as near-infrared reflectance spectroscopy or utilizing a species with a reliable relationship between oil content and seed mass (Spielbauer et al., 2009).

Further, seeds stored in the room, 4oC, and -18oC treatments did not lose viability over the 24-month storage period. This was not surprising since P50 values for various crop species ranged from 2-15 years in ambient storage and other Asteraceae species

o displayed a mean P50 of 49 years at -18.0 C (Nagel and Börner, 2010; Walters et al.,

2005). A greater storage duration is necessary to evaluate mass-dependent

106

deterioration kinetics in R. mollis in colder and drier conditions. On the contrary, viability reduced to an average of 11% across classes for seeds stored in the shed after 24 months. A positive mass-dependent germination response appeared early from 6-14 months of storage except for light class germination at month 10. Subsequent sampling months did not follow this trend and final germination percentage was highly variable.

Unfortunately, I believe the inconsistent and variable final germination data was due to the glass storage vials and not mass. The threaded plastic caps loosened with time in the shed while remaining tightly torqued in the room. Variation in cap torque and air exchange in shed vials may have caused variation in final germination percent data.

Nevertheless, I can reasonably conclude that seed mass does not promote differential deterioration in R. mollis seeds stored outside in Florida and P50 is approximately 15 months: an important consideration for native seed producers that are storing seeds in non-climate-controlled conditions for extended periods.

Relative longevity assessments performed on fresh seeds and repeated during month 0 prior to experimentation revealed a mass-independent response. This conclusion runs contrary to Genna and Pérez (2016), Guzzon et al. (2018), and Mira et al. (2019) that report mass-dependent relative longevity within Rudbeckia, Aegilops, and

Hirschfeldia accessions, respectfully. High temperature aging is purported to predict relative longevity and can highlight species or populations that are potentially short or long lived in storage (Newton et al., 2009). However, that claim has not been tested extensively on a smaller scale such as in different mass seeds within an accession.

Here, I can tentatively conclude that SSAA performed on fresh seeds did predict relative longevity in the shed, accepting that variation within the shed treatment was due to the

107

storage vials, and that seed viability was approximately 0% in all classes by month 24.

However, the observed variation minimizes the strength of this conclusion and merits repeating this study with different storage containers. A second point of criticism is this study’s use of SSAA instead of controlled deterioration. Subjecting seeds to a different combination of temperature and RH may have elicited a different response (Lehner et al., 2008) Finally, fresh seed aging revealed that seeds in all mass classes were similarly mature since longevity following high temperature aging is comparatively reduced in immature seeds (Probert et al., 2007).

Repeated relative longevity assessments revealed mass-dependent aging responses at 6 and 12 months in the shed while room stored seeds did not exhibit mass-dependent differences at any time point. Analyzing each assessment independently would have permitted false conclusions about potential mass-dependent aging in R. mollis. This evidence reduces the conclusive power of Genna and Pérez

(2016), Guzzon et al. (2018), and Mira et al. (2019) since all studies utilized a single time point. Further, seeds in Genna and Pérez (2016) were approximately two years old while accessions analyzed by Guzzon et al. (2018) were approximately 1-1.5 years old before testing. In this research, the heavy and intermediate classes with the lowest P50 values at 6 and 12 months, respectfully, were most likely inside vials with a relatively loose cap. Further evidence that this response was stochastic and not due to mass is the marked similarity in aging responses among mass classes stored in the room despite the considerable reduction in P50. Collectively, this research casts doubt on a universal relationship between intra-population seed mass variation and longevity and necessitates repetition in other species.

108

Finally, and what may be the most important take home message, is that repeated SSAA testing showed that seeds stored in the room for two years lost considerable vigor that was not revealed in final germination percent data. For example, final germination percentage was ≥ 91% for seeds stored in the room averaged across classes for the 24-month period. However, P50 following SSAA was an average of 32.2 days across classes when seeds were fresh which reduced to an average of 8.9 days when seeds were 24 months old. This highlights how P50 values generated with high temperature aging will change as seeds age and is affected by pre-storage factors.

Crawford et al. (2011) demonstrate a similar conclusion with serotinous cones of different age. As a consequence, seed longevity may be overestimated if fresh seeds are compared to old seeds and underestimated if old seeds are compared to fresh seeds in large scale interspecific studies. For example, Merritt et al. (2014) and Probert et al. (2009) mention that aging tests were conducted on seed lots with ≥ 90% and ≥

85% viability, respectfully, while Satyanti et al. (2018) do not provide pre-testing seed lot viability. Further, Probert et al. (2009) used seed lots that were between 2 and 35 years old and were previously stored at -20oC. Merritt et al. (2014) report using fresh and previously stored seeds and Satyanti et al. (2018) used only previously stored seeds.

Although seed lot viability was high in Merritt et al. (2014), this research demonstrated that older seed lots can return considerably reduced P50 values following high temperature aging. Therefore, I emphasize here that seed lot viability should be high before conducting relative longevity assessments, but more importantly, seed age should be low to obtain maximum P50 values. I also recommend that seed age should

109

accompany P50 values in relevant tables or that collection and testing dates should be provided in relevant text to inform readers.

Conclusion

Seed mass has not been extensively targeted as an explanatory factor regulating intra-accession seed longevity. Prior empirical storage studies suggest that there is no interspecific relationship between seed mass and longevity (Walters et al., 2005). This research supports that claim on an intra-accession basis for seeds stored in a non- climate-controlled shed. This research also does not support recent research suggesting that relative longevity following high temperature aging may be mass- dependent within accessions. Most importantly, this research demonstrated that seed age and pre-storage conditions have a dramatic impact on a seed’s ability to tolerate aging stress and can influence conclusions drawn from interspecific relative longevity assessments.

110

Table 4-1. Final germination percentages for Rudbeckia mollis mass classes following equilibration to various relative humidity environments. Relative humidity was converted to water potential (MPa) using the formula: 푀푃푎 = 푅∙푇 푅퐻 ln , where R = universal gas constant (8.31 J mol-1 K-1), T = absolute 푉̅푤 100 -6 3 temperature in K, 푉̅푤 is the partial molal volume of water (18.0 × 10 m mol-1), and RH = % relative humidity. Equilibration environment Final germination percentage Salt RH (%) -MPa Light Intermediate Heavy KNO3 91.0 12.9 95 89 90 KCL 85.0 22.3 95 88 94 NaCl 75.0 39.5 86 93 98 Ca(NO3)2 50.5 93.8 95 92 87 NaI 38.0 132.8 93 94 91 MgCl2 32.5 154.3 97 85 90 CaCl2 29.5 167.6 93 88 85 KAc 25.0 190.3 95 88 92 LiCl 13.0 280.1 85 87 92 KOH 8.0 346.7 96 96 92 ZnCl2 5.5 398.2 96 91 89 P2O5 0.5 797.5 88 95 95

111

Table 4-2. Deterioration parameters for Rudbeckia mollis mass classes when fresh and on month 0 of the storage experiment determined from the Ellis and Roberts viability equation. Storage duration Mass class Ki  (days) P50 (days) Fresh Light 9.14 7.35 30.4 Intermediate 8.73 8.77 32.7 Heavy 9.14 8.06 33.4 Month 0 Light 8.16 9.26 29.3 Intermediate 8.93 7.25 28.5 Heavy 8.44 8.06 27.7

112

Table 4-3. Deterioration parameters for Rudbeckia mollis mass classes determined from the Ellis and Roberts viability equation following 24 months of storage in a shed and in a room. Room Shed Storage duration Mass class Ki  (days) P50 (days) Ki  (days) P50 (days) Month 6 Light 7.78 7.81 21.7 7.46 6.94 17.1 Intermediate 7.89 7.46 21.6 7.40 8.06 19.4 Heavy 8.77 6.17 23.3 59.2 0.101 5.51 Month 12 Light 7.22 8.33 18.5 7.40 5.56 13.3 Intermediate 7.73 7.25 19.8 5.00 5.88 1.12a Heavy 7.66 7.69 20.5 7.60 6.10 15.9 Month 18 Light 7.20 5.68 12.5 60.0 0.10 5.50 Intermediate 6.65 7.69 12.7 108 0.05 5.25 Heavy 7.20 5.88 12.9 6.73 5.00 8.65 Month 24 Light 6.28 7.58 9.70 5.00 4.17 1.26a Intermediate 6.17 6.02 7.05 - - -b Heavy 6.16 8.47 9.83 - - -b a P50 approximated from curve fitting in Figure 4-5 as probit analysis yielded a P50 of 0.0. bFinal germination percentage did not exceed 50%.

113

Figure 4-1. Water sorption isotherms for Rudbeckia mollis mass classes following equilibration to various relative humidity environments. Line connecting both axes to the fitted curve indicates water content before adjustment of 6.89 ± 0.15% corresponding to a water potential of -63.2 MPa. Fitted curve is a cubic polynomial function. Mean water content values pooling mass classes were used to create a fitted equation for R. mollis water sorption: WC = 0.34 + 0.35(RH) – 0.008(RH)2 + 0.00006(RH)3.

114

Figure 4-2. Climate inside the non-climate controlled outdoor shed from January 1 to December 31, 2018. A) Air temperature and B) relative humidity.

115

Figure 4-3. Final germination percentage for Rudbeckia mollis mass classes following storage for 24 months. A) Shed (variable temperature and relative humidity (RH)), B) room (23oC, 42.5% RH), C) refrigerator (4oC, 15% RH), and D) freezer (-18oC, 15% RH). Solid line in (A) passes through average final germination percentages denoted by crosses for the light, intermediate, and heavy classes at each viability assessment. Ki and  in (A) are in probits.

116

Figure 4-4. Final germination percentages of Rudbeckia mollis mass classes subjected to saturated salt accelerated aging stress for 45 days at 41.0oC. A) Fresh and B) at month 0 of the storage experiment. Solid lines represent fitted deterioration curves.

117

Figure 4-5. Final germination percentages of Rudbeckia mollis mass classes stored in a shed and room and subjected to saturated salt accelerated aging at 41.0oC for 45 days at different time points. A) Shed, 6 months. B) Room, 6 months. C) Shed, 12 months. D) Room, 12 months. E) Shed, 18 months. F) Room, 18 months. G) Shed, 24 months. H) Room, 24 months. Solid lines represent fitted deterioration curves.

118

CHAPTER 5 CONCLUSIONS

Chapter 1 presented intraspecific seed mass variation as a driver of plant ecology and as a problem for successful seed conservation. I also outlined a theoretical framework to integrate longevity and ecological research in what I call the “general ecological model for intraspecific seed mass variation.” This model states that seed mass may dictate seed roles in nature through quantifiable effects on germination ecology and longevity. For example, research to date has shown that higher mass seeds have a germination or competitive advantage and may serve as primary recruiting individuals within populations (Moles and Westoby, 2004; Turnbull et al.,

2004). Lower mass seeds on the other hand, with a disposition for persistence

(Thompson et al., 1993), and potential greater longevity (Genna and Pérez, 2016;

Guzzon et al., 2018), may serve as placeholders in the soil seed bank. Guzzon et al.

(2018) independently proposed a similar relationship relating seed mass, germination, and longevity in Aegilops. In this final chapter, I will discuss each major hypothesis outlined in chapter 1 in the context of this proposed model and debate how my results support or refute the model’s claims.

Chapter 2 addressed the potential for mass-dependent physical and physiological differences in R. mollis seeds and how those differences may correlate with developmental location. This research was also performed using a framework of relating inherent mass-dependent differences as factors driving differential longevity studied in Chapters 3 and 4. Collectively, I was able to show that seed mass varies widely within R. mollis plants with 68.8% of variation occurring within individuals and only 30.1% occurring among individuals. Within inflorescence seed mass variation

119

accounted for the largest proportion of within individual seed mass variation totaling to

40.3%. Cellular anatomy did not differ with seed mass and major protein, starch, and lipid components appeared similar when examined visually. The most significant finding from this work was the revelation of conditional dormancy in higher mass seeds that reduced germination to 73% when exposed to fall (27/17oC) and winter (22/11oC) temperatures consistent with environmental conditions at shedding. I speculated that developmental location may drive inherent conditional dormancy at shedding in higher mass seeds since seeds of different mass develop in different locations within mother plants. For example, higher mass seeds were found to occur on higher order primary branches, lower order secondary branches, and proximal within inflorescences. Future work should evaluate this hypothesis by following individual seeds of different mass during development to 1) verify conditional dormancy when fresh, and 2) understand how conditional dormancy changes following manipulation of resource dynamics. For instance, do lower mass seeds become conditionally dormant if higher mass seeds are removed from the mother plant early during development? If resource effects are driving conditional dormancy in R. mollis, then manipulating resource dynamics should increase conditional dormancy in lower mass seeds.

In the context of my proposed model, R. mollis does not appear to behave like other species regarding mass-based germination dynamics. For example, intraspecific seed mass studies of germination performance establish that higher mass seeds germinate faster and to a higher percentage compared to lower mass seeds (see references in Chapter 1, page 12). To the contrary, fresh R. mollis seeds display an opposite relationship and non-dormant seeds display a mass-independent relationship

120

(Genna and Pérez, 2016). This suggests that seed mass regulates R. mollis germination in fall and winter following shedding. In spring and following the alleviation of inherent dormancy during winter, seed mass may play no or little role within a population of R. mollis as all seeds may have an equal probability of germinating from the soil seed bank (Chapter 3). In this context, the germination ecology of R. mollis does not support my model that is predicated on a mass-dependent germination response in non-dormant seeds. Mass-independent germination is also reported in other species and it is unclear what links these species. Perhaps mass-dependent germination is related to the formation of a persistent seed bank. I do not currently know if R. mollis forms a persistent seed bank but Chapter 3 established that seeds are capable of surviving in soil for at least two years. An in situ germination phenology study following shallow burial is necessary to address these findings.

In Chapter 3, my focus shifted away from inherent physical and physiological characteristics and moved towards addressing potential mass-based longevity during in situ burial. Current research suggests that there is a positive relationship between seed survival in soil and seed coat thickness independent of seed mass (Davis et al., 2008;

Davis et al., 2016; Gardarin et al., 2010; Schutte et al., 2014; Tricault et al., 2017). I tested that hypothesis with a two-year burial study by asking if higher mass seeds would deteriorate at a slower rate compared to lower mass seeds. I based this hypothesis on

Genna (2015) that showed pericarp thickness was greatest in heavy class R. mollis seeds. I reanalyzed that data in Chapter 3 with linear regression and showed a weak but significant positive relationship between pericarp thickness and seed mass. Chapter

3 demonstrated that R. mollis undergoes dormancy cycling during burial consistent with

121

other species that express non-deep physiological dormancy at shedding.

Unfortunately, two years was insufficient for R. mollis seeds to lose viability during burial; although the 24-month time point for the heavy class was the first to show a reduction in viability. Nevertheless, a weak but significant correlation between filled seed counts and burial months across classes indicated that higher mass seeds are more likely to be recovered filled. Therefore, seed viability did not decrease during burial but higher mass seeds were less likely to succumb to pathogen pressure. I speculate that these findings are due in part to greater seed coat thickness in higher mass seeds.

Chapter 3 results do not support my proposed model, which states that lower mass seeds have a disposition for persistence and may have greater longevity. I suspect that the lack of support is due to a multitude of abiotic and biotic factors that can reduce seed viability during burial and why the term “survival” may be more appropriate to describe long-term seed burial studies as opposed to longevity: a term most appropriate for ex situ seed storage. For example, seeds face a high but short- term predation risk from various vertebrates that decreases following burial (Hulme,

1998). Within soil, seeds face pressure from invertebrates that consume and move seeds through the soil profile, and fungi and bacteria consume viable tissue if physical seed coat barriers are compromised. Abiotic factors including temperature and soil moisture will also have a profound effect on viability loss, working in concert to reduce seed viability during periods of higher soil temperature and water potential. Seed coat thickness may be one factor that promotes survival during burial despite recent research to the contrary (Zalamea et al., 2018). Chapter 3 suggests that different mass

R. mollis seeds have a similar ability to tolerate aging stress during burial in the short

122

term but may have differing survivorship due to physical differences. A longer-term study targeting within population seed survival is needed to bolster these claims.

Finally, Chapter 4 explored potential mass-based longevity within an ex situ conservation context. I based this research on the considerable intraspecific variation in longevity documented in long-term storage studies (Walters et al., 2005). In Chapter 1, I proposed the hypothesis that seed mass variation within accessions may promote differential longevity. I based my hypothesis on two recent studies that utilized high temperature aging assays to explore longevity in different mass seeds. First, Genna and

Pérez (2016) showed that lower mass seed germination was 1.7 times greater than higher mass seeds following 20 days of aging stress. In a second study, Guzzon et al.

(2018) also showed that lower mass seeds were longer lived compared to higher mass seeds within Aegilops spikelets following controlled deterioration. Unfortunately, empirical storage research presented in Chapter 4 demonstrated no mass-dependent deterioration in seeds stored in conditions ranging from a non-climate controlled shed to

-18oC. Further, no mass-dependent germination response was evident following repeated high temperature aging over two years. I argue that this research does not support my hypothesis and that short-term seed storage does not promote mass- dependent aging in R. mollis.

In contrast to Genna and Pérez (2016) and Guzzon et al. (2018), research presented in Chapter 4 does not support my general ecological model for intraspecific seed mass variation nor does it support the similar framework presented by Guzzon et al. (2018). Unfortunately, repeated SSAA on seeds stored in the shed demonstrated marked variability that I believe was due to defective storage vials. Therefore, I do not

123

believe any meaningful mass-dependent revelations are obtainable from the shed treatment. In contrast, consistent and informative aging results were obtained from seeds stored at room temperature. Seed viability, as measured by germination percentage, did not decrease over the 24-month storage period in the room. However,

P50 for fresh seeds fell from 32.2 days to 8.9 days after 24 months of storage in the room. Germination dynamics following SSAA were also consistent across all mass classes in contrast to Genna and Pérez (2016), Guzzon et al. (2018), and Mira et al.

(2019). Inconsistencies between Chapter 4 and the above studies may have occurred for several reasons. For instance, Genna and Pérez (2016) only report one time point.

Seeds in that case were stored at 4oC for two years but were exposed to room temperature conditions multiple times. Guzzon et al. (2018) also used a single time point in their research and seeds were approximately 1-1.5 years old when subjected to controlled deterioration. Unlike Genna and Pérez (2016) however, Guzzon et al. (2018) tested multiple Aegilops species and found consistent results across accessions. It is unclear how research presented by Guzzon et al. (2018) would change if seeds were tested fresh and at multiple time points following storage. This would eliminate concerns that seed age and pre-storage factors may have affected conclusions made by Guzzon et al. (2018).

In closing, my “general ecological model for intraspecific seed mass variation” synthesizes many studies that report mass-dependent germination ecology and potential mass-dependent longevity across many different species. Unfortunately, R. mollis does not appear to fit this model since germination is not mass-dependent following the alleviation of inherent dormancy and there is no evidence of mass-

124

dependent longevity following in situ burial and ex situ storage. This research was the first to examine intra-accession seed mass variation as a source of differential seed longevity in an empirical ex situ storage study. This research also adds to the handful of studies that examined intra-population seed mass variation as a source of differential survival during burial. It is possible that other species will fit my proposed model since this topic has not been extensively studied and intra-population variation will always be masked by broad scale interspecific studies.

In the broader context, and to answer the ultimate question that this dissertation proposes, do seeds of different mass within populations die at different rates? In the context of in situ burial, the answer is clearly yes although overwhelming support for this claim is lacking in the literature. I draw this conclusion from the fact that seeds die in soil due to numerous abiotic and biotic factors. A seed’s physical and chemical composition deters biotic factors during burial while a seeds inherent ability to resist physiological aging determine its resistance to abiotic factors including temperature and soil moisture.

Biotic factors play a much larger role in regulating seed survival during burial and cases of seeds surviving for centuries in soil are rare. The relationship between seed mass and survival during burial will not always be positive, as there are reported instances of negative relationships between seed mass, seed coat thickness, and survival (Schutte et al., 2014). Therefore, it is not seed mass per se that drives survival during burial.

Rather, it is the species-specific physical and chemical factors that correlate with seed mass within a population.

In the context of ex situ storage, there is currently no evidence proposed by this research or by other empirical storage studies that would suggest a relationship exists

125

between interspecific, intraspecific, or intra-accession seed mass variation and longevity. The few studies to date that suggest a relationship exists, either interspecific or intraspecific, are a product of high temperature aging (Genna and Pérez, 2016;

Guzzon et al., 2018; Merritt et al., 2014, Mira et al., 2019; Satyanti et al., 2018). For that reason, empirical storage studies are always necessary to verify claims purported by high temperature aging, particularly when high temperature aging is utilized as a proxy for empirical storage.

126

LIST OF REFERENCES

Afonso, A., Castro, S., Loureiro, J., Mota, L., Cerca de Oliveira, J., and Torices, R. (2014) The effects of achene type and germination time on plant performance in the heterocarpic Anacyclus clavatus (Asteraceae). American Journal of Botany 101, 892-898.

Alkio, M., Schubert, A., Diepenbrock, W., and Grimm, E. (2003) Effect of source-sink ratio on seed set and filling in sunflower (Helianthus annuus L.). Plant, Cell and Environment 26, 1609-1619.

AOSA (2002) Seed vigor testing handbook: Contribution No.32 to the handbook on seed testing. Association of Official Seed Analysts, Washington, DC, USA.

Ballesteros, D. and Walters, C. (2011) Detailed characterization of mechanical properties and molecular mobility within dry seed glasses: relevance to the physiology of dry biological systems. The Plant Journal 68, 607-619.

Ballesteros, D., Hill, L.M., Lynch, R.T., Pritchard, H.W., and Walters, C. (2018) Longevity of preserved germplasm: The temperature dependency of aging reactions in glassy matrices of dried fern spores. Plant and Cell Physiology 60, 376-392.

Baloch, H.A., DiTommaso, A., and Watson, A.K. (2001) Intrapopulation variation in Abutilon theophrasti seed mass and its relationship to seed germinability. Seed Science Research 11, 335-343.

Banovetz, S.J. and Scheiner, S.M. (1994) The effects of seed mass on the seed ecology of Coreopsis lanceolata. The American Midland Naturalist 131, 65-74.

Baskin, C.C. and Baskin, J.M. (2014) Seeds: ecology, biogeography, and evolution of dormancy and germination. Academic Press, San Diego, CA.

Bramble, T., Herrman, T.J., Loughin, T., and Dowell, F. (2002) Single Kernel Protein variance structure in commercial wheat fields in western Kansas. Crop Science 42, 1488-1492.

Buitink, J. and Leprince, O. (2004) Glass formation in plant anhydrobiotes: survival in the dry state. Cryobiology 48, 215-228.

Buitink, J. and Leprince, O. (2008) Intracellular glasses and seed survival in the dry state. Biologies 331, 788-795.

Calderini, D.F. and Ortiz-Monasterio, I. (2003) Grain position affects grain macronutrient and micronutrient concentrations in wheat. Crop Science 43, 141-151.

127

Castañeda-Álvarez, N.P., Khoury, C.K., Achicanoy, H.A., Bernau, V., Dempewolf, H., Eastwood, R.J., Guarino, L., Harker, R.H., Jarvis, A., Maxted, N., Müller, J.V., Ramirez-Villegas, J., Sosa, C.C., Struik, P.C., Vincent, H., and Toll, J. (2016) Global conservation priorities for crop wild relatives. Nature Plants 2, 1-6.

Chee-Sanford, J.C., Williams, M.M., Davis, A.S., and Sims, G.K. (2006) Do microorganisms influence seed-bank dynamics? Weed Science 54, 575-587.

Conn, J.S., Beattie, K.L., and Blanchard, A. (2006) Seed viability and dormancy of 17 weed species after 19.7 years of burial in Alaska. Weed Science 54, 464-470.

Corbineau, F., Gay-Mathieu, C., Vinel, D., and Côme, D. (2002) Decrease in sunflower (Helianthus annuus) seed viability caused by high temperature as related to energy metabolism, membrane damage and lipid composition. Physiologia Plantarum 116, 489-496.

Crawford, A.D., Plummer, J.A., Probert, R.J., and Steadman, K.J. (2011) The influence of cone age on the relative longevity of Banksia seeds. Annals of Botany 107, 303-309.

Crochemore, M.L., Huyghe, C., Papineau, J., and Julier, B. (1994) Intra-plant variability in seed size and seed quality in Lupinus albus L. Agronomie 14, 5-13.

Dai, C., Li, L., Wang, Z., and Liao, K. (2018) Sequential decline in fruit resource allocation within inflorescences of Sagittaria trifolia: a test of non-uniform pollination hypothesis. Plant Species Biology 33, 259-267.

Dalling, J.W., Davis, A.S., Schutte, B.J., and Arnold, A.E. (2011) Seed survival in soil: interacting effects of predation, dormancy and the soil microbial community. Journal of Ecology 99, 89-95.

Davis, A.S., Schutte, B.J., Iannuzzi, J., and Renner, K.A. (2008) Chemical and physical defense of weed seeds in relation to soil seedbank persistence. Weed Science 56, 676-684.

Davis, A.S., Fu, X., Schutte, B.J., Berhow, M.A., and Dalling, J.W. (2016) Interspecific variation in persistence of buried weed seeds follows trade-offs among physiological, chemical, and physical seed defenses. Ecology and Evolution 6, 6836-6845.

Delouche, J.C. and Baskin, C.C. (1973) Accelerated aging techniques for predicting the relative storability of seed lots. Seed Science and Technology 1, 427-452.

Diggle, P.K. (1995) Architectural effects and the interpretation of patterns of fruit and seed development. Annual Review of Ecological Systems 26, 531-532.

128

Dwivedi, S.L., Jambunathan, R., and Nigam, S.N. (1990) Relationship of seed mass to oil and protein contents in peanut (Arachis hypogaea L.). Peanut Science 17, 48- 52.

Ellis, R. (2019) Temporal patterns of seed quality development, decline, and timing of maximum quality during seed development and maturation. Seed Science Research, 1-8.

Ellis, R.H and Roberts, E.H. (1980) Improved equations for the prediction of seed longevity. Annals of Botany 45, 13-30.

El-Maarouf-Bouteau, H., Mazuy, C., Corbineau, F., and Bailly, C. (2011) DNA alteration and programmed cell death during ageing of sunflower seed. Journal of Experimental Botany 62, 5003-5011.

Espinosa-Garcia, F.J., Vazquez-Bravo, R., and Martinez-Ramos, M. (2003) Survival, germinability and fungal colonization of dimorphic achenes of the annual weed Galinsoga parviflora buried in the soil. Weed Research 43, 269-275.

FAO (2014) Genebank standards for plant genetic resources for food and agriculture. Revised edition. Rome.

Gardarin, A., Durr, C., Mannino, M.R., Busset, H., and Colbach, N. (2010) Seed mortality in the soil is related to seed coat thickness. Seed Science Research 20, 243-256.

Genna, N.G. (2015) Seed mass influences germination and viability of thermally stressed wildflower seeds. M.S. thesis, University of Florida, Gainesville, FL.

Genna, N.G. and H.E. Pérez. (2016) Mass-based germination dynamics of Rudbeckia mollis (Asteraceae) seeds following thermal and aging stress. Seed Science Research 26, 231-244.

Gremer, J.R. and Venable, D.L. (2014) Bet hedging in desert winter annual plants: optimal germination strategies in a variable environment. Ecology Letters 17, 380-387.

Groot, S.P.C., Surki, A.A., de Vos, R.C.H., and Kodde, J. (2012) Seed storage at elevated partial pressure of oxygen, a fast method for analysing seed ageing under dry conditions. Annals of Botany 110, 1149-1159.

Guillemin, J. and Chauvel, B. (2011) Effects of the seed weight and burial depth on the seed behavior of common ragweed (Ambrosia artemisiifolia). Weed Biology and Management 11, 217-223.

129

Guleria, S., Sharma, S., Gill, B.S., and Munshi, S.K. (2008) Distribution and biochemical composition of large and small seeds of soybean (Glycine max L.). Journal of the Science of Food and Agriculture 88, 269-272.

Guzzon, F., Orsenigo, S., Gianella, M., Muller, J.V., Vagge, I., Rossi, G., and Mondoni, A. (2018) Seed heteromorphy influences seed longevity in Aegilops. Seed Science Research 28, 277-285.

Harper, J.L., Lovell, P.H., and Moore, K.G. (1970) The shapes and sizes of seeds. Annual Review of Ecology and Systematics 1, 327-356.

Hassan, F., Kaleem, S., and Ahmad, M. (2011) Oil and fatty acid distribution in different circles of sunflower head. Food Chemistry 128, 590-595.

Hay, F.R. and Probert, R.J. (2011) Collecting and handling seeds in the field. In: Guarino, L., Ramanatha Rao, V., and Reid, R. (eds.) Collecting plant genetic diversity: technical guidelines – 2011 update. CAB International, Wallingford, UK.

Hay, F.R. and Probert, R.J. (2013) Advances in seed conservation of wild plant species: a review of recent research. Conservation Physiology 1, 1-12.

Hendrix, S.D. (1984) Variation in seed weight and its effects on germination in Pastinaca sativa L. (Umberlliferae). American Journal of Botany 71, 795-802.

Hendrix, S.D. and Sun, I. (1989) Inter- and intraspecific variation in seed mass in seven species of umbellifer. New Phytologist 112, 445-451.

Hendrix, S.D. and Trapp, J.E. (1992) Population demography of Pastinaca sativa (Apiaceae): Effects of seed mass on emergence, survival, and recruitment. American Journal of Botany 79, 365-375.

Hernandez, L.F. and Orioli, G.A. (1985) Imbibition and germination rates of sunflower (Helianthus annus L.) seeds according to fruit size. Field Crops Research 10, 355-360.

Hoekstra, F.A., Golovina, E.A., and Buitink, J. (2001) Mechanisms of plant desiccation tolerance. Trends in Plant Science 6, 431-438.

Hulme, P.E. (1998) Post-dispersal seed predation: consequences for plant demography and evolution. Perspectives in Plant Ecology, Evolution and Systematics 1, 32- 46.

Hou, J. and Romo, J.T. (1998) Seed weight and germination time affect growth of 2 shrubs. Journal of Range Management 51, 699-703.

130

Jianhua, Z. and McDonald, M.B. (1996) The saturated salt accelerated aging test for small-seeded crops. Seed Science and Technology 25, 123-131.

Kaeser, M. and Kirkman, K. (2012) Seed longevity of 12 native herbaceous species in a fire-maintained pine savanna after 8 years of burial. Forest Ecology and Management 281, 68-74.

Kaleem, S., Hassan, F.U., Bukhsh, M.A.A.H.A, Mahmood, I., Ullah, R., Ahmad, M., and Wasaya, A. (2011) Oil and oil quality in different circles of mature sunflower head as influenced by varying environments. Pakistan Journal of Nutrition 10, 373-377.

Kettner, K. and Pérez, H.E. (2012) Dose-response of germinating Rudbeckia mollis (Asteraceae) seeds exposed to various thermal scenarios. Seed Science Research 22, 191-197.

Khan, M.L. (2004) Effects of seed mass on seedling success in Artocarpus heterophyllous L., a tropical tree species of north-east India. Acta Oecologica 25, 103-110.

Khan, M.A. and Ungar, I.A. (1984) The effect of salinity and temperature on the germination of polymorphic seeds and growth of Atriplex triangularis Willd. American Journal of Botany 71, 481-489.

Kilber, A. and Eckert, C.G. (2004) Sequential decline in allocation among flowers within inflorescences: Proximate mechanisms and adaptive significance. Ecology 85, 1675-7687.

Kochanek, J., Steadman, K.J., Probert, R.J., and Adkins, S.W. (2009) Variation in seed longevity among different populations, species and genera found in collections from wild Australian plants. Australian Journal of Botany 57, 123-131.

Kolodziejek, J. (2017) Effect of seed position and soil nutrients on seed mass, germination and seedling growth in Peucedanum oreoselinum (Apiaceae). Scientific Reports 7, 1-11.

Kranner, I., Minibayeva, F.V., Beckett, R.P., and Seal, C.E. (2010) What is stress? Concepts, definitions and applications in seed science. New Phytologist, 188, 655-673.

Krannitz, P.G. (1997) Variation in magnesium and nitrogen content in seeds of antelope bitterbrush (Purshia tridentata, Rosaceae). American Journal of Botany 84, 1738- 1742.

Larsen, S.U. and Andreasen, C. (2004) Light and heavy turfgrass seeds differ in germination percentage and mean germination thermal time. Crop Science 44, 1710-1720.

131

Lehner, A., Mamadou, N., Poels, P., Côme, D., Bailly, C., and Cornineau, F. (2008) Changes in soluble carbohydrates, lipid peroxidation and antioxidant enzyme activities in the embryo during ageing in wheat grains. Journal of Cereal Science 47, 555-565.

Lehtila, K. and Ehrlen, J. (2005) Seed Size as an indicator of seed quality: A case study of Primula veris. Acta Oecologica 28, 207-212.

Leprince, O., Pellizzaro, A., Berriri, S., and Buitink, J. (2017) Late seed maturation: drying without dying. Journal of Experimental Botany 68, 827-841.

Long, R.L., Panetta, F.D., Steadman, K.J., Probert, R., Bekker, R.M., Brooks, S., and Adkins, S.W. (2008) Seed persistence in the field may be predicted by laboratory-controlled aging. Weed Science 56, 523-528.

Long, R.L., Kranner, I., Panetta, F.D., Birtic, S., Adkins, S.W., and Steadman, K.J. (2011) Wet-dry cycling extends seed persistence by re-instating antioxidant capacity. Plant and Soil 338, 511-519.

Long, R.L., Gorecki, M.J., Renton, M., Scott, J.K., Colville, L., Goggin, D.E., Commander, L.E., Westcott, D.A., Cherry, H., and Finch-Savage, W.E. (2015) The ecophysiology of seed persistence: a mechanistic view of the journey to germination or demise. Biological Reviews 90, 31-59.

Lönnberg, K. and Eriksson, O. (2013) Rules of the seed size game: contests between large-seeded and small-seeded species. Oikos 122, 1080-1084.

Marshall, D.L. (1986) Effect of seed size on seedling success in three species of Sesbania (Fabaceae). American Journal of Botany 73, 457-464.

Masin, R., Zuin, M.C., Otto, S., and Zanin, G. (2006). Seed longevity and dormancy of four summer annual grass weeds in turf. Weed Research 46, 362-370.

Matthews, S., Noli, E., Demir, I., Khajeh-Hosseini, M., and Wagner, M.H. (2012) Evaluation of seed quality: from physiology to international standardization. Seed Science Research 22, S69-S73.

McGinley, M.A., Temme, D.H., and Geber, M.A. (1987) Parental investment in offspring in variable environments: Theoretical and empirical considerations. The American Naturalist 130, 370-398.

Merritt, D.J., Martyn, A.J., Ainsley, P., Young, R.E., Seed, L.U., Thorpe, M., Hay, F.R., Commander, L.E., Shackelford, N., Offord, C.A., Dixon, K.W., and Probert, R.J. (2014) A continental-scale study of seed lifespan in experimental storage

132

examining seed, plant, and environmental traits associated with longevity. Biodiversity Conservation 23, 1081-1104.

Michaels, H.J., Benner, B., Hartgerink, A.P., Lee, T.D., and Rice, S. (1988) Seed size variation: magnitude, distribution, and ecological correlates. Evolutionary Ecology 2, 157-166.

Milberg, P., Andersson, L., Elfverson, C., and Regner, S. (1996) Germination characteristics of seeds differing in mass. Seed Science Research 6, 191-198.

Mira, S., Veiga-Barbosa, L., and Pérez-García, F. (2019) Seed dormancy and longevity variability of Hirschfeldia incana L. during storage. Seed Science Research, 1-7.

Moles, A.T., Warton, D.I., and Westoby, M. (2003) Seed size and survival in the soil in arid Australia. Austral Ecology 28, 575-585.

Moles, A.T. and Westoby, M. (2004) Seedling survival and seed size: a synthesis of the literature. Journal of Ecology 92, 372-383.

Moravcova, L., Perglova, I., Pysek, P., Jarosik, V., and Pergl, J. (2005) Effects of fruit position on fruit mass and seed germination in the alien species Heracleum mantegazzianum (Apiaceae) and the implications for its invasion. Acta Oecologica 28, 1-10.

Mounce, R., Smith. P., and Brockington, S. (2017) Ex situ conservation of plant diversity in the world’s botanic gardens. Nature Plants 3, 795-802.

Munshi, S.K., Kaushal, B., and Bajaj, R.K. (2003) Composition changes in seeds influenced by their positions in different whorls of mature sunflower head. Journal of the Science of Food and Agriculture 83, 1622-1626.

Murdoch, A.J. and Ellis, R.H. (2000) Dormancy, viability, and longevity, p. 183-214. In: Fenner, M. (ed.). Seeds: The ecology of regeneration in plant communities. 2nd ed. CABI, New York, NY, USA.

Nagel, M. and Börner, A. (2010) The longevity of crop seeds stored under ambient conditions. Seed Science Research 20, 1-12.

Nagel, M., Kodde, J., Pistrick, S., Mascher, M., Börner, A. and Groot, S.P.C. (2016) Barely seed aging: genetics behind the dry elevated pressure of oxygen aging and moist controlled deterioration. Frontiers in Plant Science 7, 1-11.

Nelson, D.M. and Johnson, C.D. (1983) Stabilizing selection on seed size in Astragalus (Leguminosae) due to differential predation and differential germination. Journal of the Kansas Entomological Society 56, 169-174.

133

Newton, R., Hay, F., and Probert, R. (2009) Protocol for comparative seed longevity testing. Royal Botanic Gardens, Kew, London, UK.

Norcini, J.G. and Aldrich, J.H. (2007) Performance of native Florida plants under north Florida conditions. Environmental Horticulture Department, UF/IFAS Extension, ENH1074.

Obeso, J.R. (1993) Seed mass variation in the perennial herb Asphodelus albus: sources of variation and position effect. Oecologia 93, 571-575.

Obeso, J.R. (2012) Mineral nutrient stoichiometric variability in Hedera helix (Araliaceae) seeds. Annals of Botany 109, 801-806.

Parker, W.C., Noland, T.L., and Morneault, A.E. (2006) The effects of seed mass on germination, seedling emergence, and early growth of eastern white pine (Pinus strobus L.). New Forests 32, 33-49.

Probert, R.J. (2003) Seed viability under ambient conditions, and the importance of drying. In: Smith, R.D., Dickie, J.B., Linington, S.H, Pritchard, H.W., and Probert, R.J. (eds.) Seed conservation: turning science into practice, The Royal Botanic Gardens, Kew, UK.

Probert, R.J and Linington, S.H. (2006) Viability, p. 740. In: Black, M., Bewley, J.D, and Halmer, P. (eds.) The encyclopedia of seeds: Science, technology and uses. CABI, Wallingford, UK.

Probert, R.J., Adams, J., Coneybeer, J., Crawford, A., and Hay, F. (2007) Seed quality for conservation is critically affected by pre-storage factors. Australian Journal of Botany 55, 326-335.

Probert, R.J., Daws, M.I., and Hay, F.R. (2009) Ecological correlates of ex situ seed longevity: a comparative study on 195 species. Annals of Botany 104, 57-69.

Razavi, S.M.A. and Milani, E. (2006) Some physical properties of the watermelon seeds. African Journal of Agricultural Research 1, 65-69.

Righter, F.I. (1945) Pinus: The relationship of seed size and seedling size to inherent vigor. Journal of Forestry 43, 131-137.

Satyanti, A., Nicotra, A.B., Merkling, T., and Guja, L.K. (2018) Seed mass and elevation explain variation in seed longevity of Australian alpine species. Seed Science Research 28, 319-331.

Schaal, B.A. (1980) Reproductive capacity and seed size in Lupinus texensis. American Journal of Botany 67, 703-709.

134

Schoeman, J., Buckley, Y.M., Cherry, H., Long, R.L., and Steadman, K.J. (2010) Inter- population variation in seed longevity for two invasive weeds: Chrysanthemoides monilifera ssp. monilifera (boneseed) and ssp. rotundata (bitou bush). Weed Research 50, 67-75.

Schutte, B.J., Davis, A.S., Peinado Jr., S.A., and Ashigh, J. (2014) Seed-coat thickness data clarify seed size-seed bank persistence tradeoffs in Abutilon theophrasti (Malvaceae). Seed Science Research 24, 119-131.

Schwember, A.R. and Bradford, K.J. (2010) Quantitative trait loci associated with longevity of lettuce seeds under conventional and controlled deterioration storage conditions. Journal of Experimental Botany 61, 4423-4436.

Schwienbacher, E., Marcante, S., and Erschbamer, B. (2010) Alpine species seed longevity in the soil in relation to seed size and shape – A 5-year burial experiment in the Central Alps. Flora 205, 19-25.

Smith, C.C. and Fretwell, S.D. (1974) The optimal balance between size and number of offspring. The American Naturalist 108, 499-506.

Spielbauer, G., Armstrong, P., Baier, J.W., Allen, W.B., Richardson, K., Shen, B., and Settles, A.M. (2009) High-throughput near-infrared reflectance spectroscopy for predicting quantitative and qualitative composition phenotypes of individual maize kernels. Cereal Chemistry 86, 556-564.

Stanton, M.L. (1984) Seed variation in wild radish: effect of seed size on components of seedling and adult fitness. Ecology 65, 1105-1112.

Stephenson, A.G. (1981) Flower and fruit abortion: Proximate causes and ultimate functions. Annual Review of Ecology and Systematics 12, 253-279.

Stokes, M.E., Davis, C.S., and Koch, G.G. (2012) Categorical data analysis using SAS, third edition. SAS Institute, Cary, NC, USA.

Thompson, K. (2000). The functional ecology of soil seed banks, p. 215-235. In: Fenner, M. (ed.). Seeds: The ecology of regeneration in plant communities. 2nd ed. CABI, New York, NY, USA.

Thompson, K., Band, S.R., and Hodgson, J.G. (1993) Seed size and shape predict persistence in soil. Functional Ecology 7, 236-241.

Thompson, K., Cerian, R.M., Bakker, J.P., and Bekker, R.M. (2003) Are seed dormancy and persistence in soil related? Seed Science Research 13, 97-100.

135

Tielbörger, K. Petrů, M., and Lampei, C. (2012) Bet-hedging germination in annual plants: a sound empirical test of the theoretical foundations. Oikos 121, 1860- 1868.

Torices, R. and Mendez, M. (2010) Fruit size decline from the margin to the center of the capitula is the result of resource competition and architectural constraints. Oecologia 164, 949-958.

Torices, R., Agudo, A., and Alvarez, I. (2013) Not only size matters: achene morphology affects time of seedling emergence in three heterocarpic species of Anacyclus (Anthemideae, Asteraceae). Anales del Jardin Botanico de Madrid 70, 48-55.

Tricault, Y., Matejicek, A., Darmency, H. (2017) Variation of seed dormancy and longevity in Raphanus raphanistrum L. Seed Science Research 28, 34-40.

Tripathi, R.S. and Khan, M.L. (1990) Seed weight and microsite characteristics on germination and seedling fitness in two species of Quercus in a subtropical wet hill forest. Oikos 57, 289-296.

Turnbull, L.A., Rees, M., and Crawley, M.J. (1999) Seed mass and the competition/colonization trade-off: a sowing experiment. Journal of Ecology 87, 899-912.

Turnbull, L.A., Coomes, D., Hector, A., and Rees, M. (2004) Seed mass and the competition/colonization trade-off: competitive interactions and spatial patterns in a guild of annual plants. Journal of Ecology 92, 97-109. van Molken, T., Jorritsma-Wienk, L.D., van Hoek, P.H.W., and de Kroon, H. (2005) Only seed size matters for germination in different populations of the dimorphic Tragopogon pratensis subsp. pratensis (Asteraceae). American Journal of Botany 92, 432-437. van Mourik, T.A., Stomph, T.J., and Murdoch, A.J. (2005) Why high seed densities within buried bags may overestimate depletion rates of soil seed banks. Journal of Applied Ecology 42, 299-305.

Vaughton, G. and Ramsey, M. (1997) Seed mass variation in the shrub Banksia spinulosa (Proteaceae): Resource constraints and pollen source effects. International Journal of Plant Sciences 158, 424-431.

Vertucci, C.W. and Roos, E.E. (1990) Theoretical basis of protocols for seed storage. Plant Physiology 94, 1019-1023.

Wagner, M. and Mitschunas, N. (2008) Fungal effects on seed bank persistence and potential applications in weed biocontrol: A review. Basic and Applied Ecology 9, 191-203.

136

Walck, J.L., Baskin, J.M., Baskin, C.C., and Hidayati, S.N. (2005) Defining transient and persistent seed banks in species with pronounced seasonal dormancy and germination patterns. Seed Science Research 15, 189-196.

Walters, C. (2005) Dying while dry: Kinetics and mechanisms of deterioration in desiccated organisms. Integrative and Comparative Biology 45, 751-758.

Walters, C. (2014) Extreme biology: Probing life at low water contents and temperatures. Acta Horticulturae 1039, 49-56.

Walters, C. (2015) Genebanking seeds from natural populations. Natural Areas Journal 35, 98-105.

Walters, C., Farrant, J.M., Pammenter, N.W., and Berjak, P. (2002) Desiccation stress and damage. p. 263-291. In: Black, M. and Pritchard, H.W. (eds.). Desiccation and survival in plants: drying without dying. CABI publishing, Wallingford, UK.

Walters, C., Wheeler, L., and Stanwood, P.C. (2004) Longevity of cryogenically stored seeds. Cryobiology 48, 229-244.

Walters, C. Wheeler, L.M., and Grotenhuis, J.M. (2005) Longevity of seeds stored in a genebank: species characteristics. Seed Science Research 15, 1-20.

Walters, C. and Koster, K.L. (2007) Structural dynamics and desiccation damage in plant reproductive organs. p. 251-279. In: Jenks, M.A. and Wood, A.J. (eds.). Plant desiccation tolerance. Blackwell Publishing, Ames, Iowa, USA.

Walters, C., Ballesteros, D., and Vertucci, V.A. (2010) Structural mechanics of seed deterioration: Standing the test of time. Plant Science 179, 565-573.

Wang, A.B., Tan, D.Y., Baskin, C.C., and Baskin, J.M. (2010) Effect of seed position in spikelet on life history of Eremopyrum distans (Poaceae) from the cold desert of north-west China. Annals of Botany 106, 95-105.

Weis, M.I. (1982) The effects of propagule size on germination and seedling growth in Mirabilis hirsuta. Canadian Journal of Botany 60, 1868-1874.

Westoby, M., Leishman, M., and Lord, J. (1996) Comparative ecology of seed size and dispersal. Philosophical Transactions of the Royal Society B 351, 1309-1318.

Winn, A.A. (1991) Proximate and ultimate sources of within-individual variation in seed mass in Prunella vulgaris (Lamiaceae). American Journal of Botany 78, 838-844.

137

Wojtyla, L., Lechowska, K., Kubala, S., and Garnczarska, M. (2016) Molecular processes induced in primed seeds – increasing the potential to stabilize crop yields under drought conditions. Journal of Plant Physiology 203, 116-126.

Wulff, R. (1973) Intrapopulational variation in the germination of seeds in Hyptis suaveolens. Ecology 54, 646-649.

Wunderlin, R.P. and Hansen, B.F. (2011) Guide to the vascular plants of Florida. University Press of Florida, Gainesville, FL.

Zalamea, P., Dalling, J.W., Sarmiento, C., Arnold, A.E., Delevich, C., Berhow, M.A., Ndobegang, A., Gripenberg, S., and Davis, A.S. (2018) Dormancy-defense syndromes and tradeoffs between physical and chemical defenses in seeds of pioneer species. Ecology 99, 1988-1998.

Zimmerman, J.K. and Weis, M.I. (1983) Fruit size variation and its effects on germination and seedling growth in Xanthium strumarium. Canadian Journal of Botany 61, 2309-2315.

138

BIOGRAPHICAL SKETCH

Nicholas Gerard Genna was born and raised in Miami, Florida. Nicholas attended the University of Florida following high school and majored in biology as an undergraduate. Nicholas’ interest in research grew into an undergraduate honors thesis project studying the efficacy of a biocide to control fungal contamination in Uniola paniculata seeds under the advisement of Dr. Héctor E. Pérez. Nicholas’ published his honors thesis in Seed Science and Technology in 2015. Nicholas continued his research prowess as a master’s student at the University of Florida with Dr. Pérez as his advisor. Nicholas studied seed mass variation effects on germination and high temperature aging in Rudbeckia mollis. Nicholas published his thesis research in Seed

Science Research in 2016. Nicholas continued his education as a doctoral student at the University of Florida with Dr. Pérez as his advisor. Nicholas received his doctorate in horticultural sciences in August 2019.

139