Mississippi State University Scholars Junction

Theses and Dissertations Theses and Dissertations

8-6-2021

Regeneration potential and habitat suitability modeling of three imperiled Southeastern U.S. woody plants

Clayton Warren Hale [email protected]

Follow this and additional works at: https://scholarsjunction.msstate.edu/td

Recommended Citation Hale, Clayton Warren, "Regeneration potential and habitat suitability modeling of three imperiled Southeastern U.S. woody plants" (2021). Theses and Dissertations. 5192. https://scholarsjunction.msstate.edu/td/5192

This Graduate Thesis - Open Access is brought to you for free and open access by the Theses and Dissertations at Scholars Junction. It has been accepted for inclusion in Theses and Dissertations by an authorized administrator of Scholars Junction. For more information, please contact [email protected]. Template A v4.3 (beta): Created by T. Robinson (01/2021)

Regeneration potential and habitat suitability modeling of

three imperiled Southeastern U.S. woody plants

By TITLE PAGE Clayton Warren Hale

Approved by:

Joshua J. Granger (Major Professor) Sandra B. Correa Janice L. DuBien Courtney M. Siegert Mohammad Sepehrifar Donald L. Grebner (Graduate Coordinator) Loren W. Burger (Dean, College of Forest Resources)

A Thesis Submitted to the Faculty of Mississippi State University in Partial Fulfillment of the Requirements for the Degree of Master of Science in Forestry in the Department of Forestry

Mississippi State, Mississippi

August 2021

Copyright by COPYRIGHT PAGE Clayton Warren Hale

2021

Name: Clayton Warren Hale ABSTRACT Date of Degree: August 6, 2021

Institution: Mississippi State University

Major Field: Forestry

Major Professor: Joshua J. Granger

Title of Study: Regeneration potential and habitat suitability modeling of three imperiled Southeastern U.S. woody plants

Pages in Study 93

Candidate for Degree of Master of Science

Presented within this thesis are three studies on three rare and imperiled Southeastern woody plant species: mountain stewartia (Stewartia ovata), Atlantic white-cedar

(Chamaecyparis thyoides), and Miller’s witch-alder (Fothergilla milleri). This work contributes to the ecological understanding of these three species allowing for better-informed conservation decision-making. Machine learning habitat suitability models are presented for mountain stewartia and Miller’s witch-alder. These models can direct limited conservation dollars and manpower towards areas of the highest habitat suitability. This work also utilizes field-based data to assess the habitat needs, species associations, and regeneration potential of both Atlantic white-cedar and Miller’s witch-alder. Understanding the habitat and regeneration potential of these species allows conservationists to make more tailored land management decisions for the species. As plant species continue to be threatened with extinction, more basic and applied research is needed to lessen the impacts of the 6th mass extinction on native flora.

DEDICATION

This thesis is dedicated to those who instilled in me from birth the importance of education and an appreciation for the natural world, my parents, John and Krista Hale, and my grandparents, Bill and Bonnie Warren and Patricia Hale. If it were not for their love, support, and encouragement I could not have possibly completed this thesis. I would also like to recognize my partner, Morgan Tate, for her love and support throughout this journey.

ii

ACKNOWLEDGEMENTS

I would first like to thank my major professor, Dr. Joshua J. Granger, for giving me the opportunity to come to Mississippi State University and pursue this research. His guidance and advice proved invaluable throughout this process. I would also like to thank my committee, Dr.

Sandra Correa, Dr. Janice DuBien, and Dr. Courtney Siegert for their support and guidance during my time here at Mississippi State. This work would not have been possible without the patience and guidance of Dr. Ali Paulson and Dr. Carlos Ramirez-Reyes. I would also like to thank the faculty of the Department of Forestry as a whole. I am forever grateful for the support provided by my fellow graduate students. Without their assistance in the office and field this would not be possible. Therefore, I would like to thank Caleb Goldsmith, Darcey Collins, Will

Kruckeberg, and Spenser Madden. I have included acknowledgement sections at the end of each chapter to thank those that assisted with the specific chapters.

iii

TABLE OF CONTENTS

DEDICATION ...... ii

ACKNOWLEDGEMENTS ...... iii

LIST OF TABLES ...... vi

LIST OF FIGURES ...... vii

CHAPTER

I. INTRODUCTION ...... 1

Literature Cited ...... 4

II. MODELING HABITAT SUITABILITY FOR STEWARTIA OVATA ACROSS THE SOUTHEASTERN UNITED STATES ...... 5

Abstract ...... 5 Introduction ...... 6 Methods ...... 8 Study Area ...... 8 Species Occurrence Data ...... 8 Environmental Predictor Data ...... 9 Maxent Model ...... 10 Results ...... 12 Discussion ...... 13 Conclusions ...... 17 Acknowledgements ...... 17 Tables ...... 18 Figures ...... 21 Literature Cited ...... 23

III. ATLANTIC WHITE-CEDAR (CHAMAECYPARIS THYOIDES) RESPONSE POST- HURRICANE DISTURBANCE...... 28

Abstract ...... 28 Introduction ...... 29 Methods ...... 32 2006 Post- Katrina Data Source ...... 32 iv

Study Site ...... 33 Atlantic White-Cedar Inventory ...... 34 Woody Plant Community Composition ...... 35 Results ...... 36 Atlantic White-Cedar Inventory ...... 36 Woody Plant Community Composition ...... 37 Discussion ...... 38 Conclusions ...... 43 Acknowledgements ...... 44 Tables ...... 45 Figures ...... 46 Literature Cited ...... 52

IV. MICROSITE HABITAT, SPECIES ASSOCIATIONS, AND HABITAT SUITABILITY MODEL OF A GLOBALLY IMPERILED SHRUB...... 56

Abstract ...... 56 Introduction ...... 57 Methods ...... 59 Sampling Areas ...... 59 Stem Counts ...... 59 Woody Vegetation Sampling ...... 60 Soil Sampling ...... 62 Habitat Suitability Model (HSM) ...... 63 Results ...... 66 Stem Counts ...... 66 Woody Vegetation Sampling ...... 66 Soil Sampling ...... 68 Habitat Suitability Model (HSM) ...... 69 Discussion ...... 70 Conclusions ...... 75 Acknowledgments ...... 76 Tables ...... 77 Figures ...... 80 Literature Cited ...... 86

V. CONCLUSION ...... 90

Literature Cited ...... 93

v

LIST OF TABLES

Table 2.1 Contribution of environmental variables to mountain stewartia habitat suitability model in the Southeastern United States...... 18

Table 2.2 Physiographic sub-ecoregions with the greatest total area identified within the top bin of habitat suitability model for mountain stewartia in the southeastern United States ...... 19

Table 2.3 Watersheds with the greatest total area identified within the top bin of habitat suitability model for mountain stewartia in the Southeastern United States ...... 20

Table 3.1 Age at breast height of Atlantic white-cedar stems within the study site ...... 45

Table 4.1 Associated woody plant species within Miller’s witch-alder populations in Geneva County (Alabama), Okaloosa County (Florida), and Baldwin County (Alabama) ...... 77

Table 4.2 Soil characteristics of the three sampled Miller’s witch-alder populations in Geneva County (Alabama), Okaloosa County (Florida), and Baldwin County (Alabama) ...... 78

Table 4.3 Contribution of environmental variables to Miller’s witch-alder habitat suitability model ...... 79

vi

LIST OF FIGURES

Figure 2.1 Flower and form of mountain stewartia in Southern Appalachia, United States...... 21

Figure 2.2 Habitat suitability model developed for mountain stewartia in the Southeastern United States ...... 22

Figure 3.1 Location of Atlantic white-cedar study site in Southern Mississippi, United States...... 46

Figure 3.2 Map of Atlantic white-cedar study site in Southern Mississippi, United States ...... 47

Figure 3.3 Locations of all Atlantic white-cedar stems found within the study site in Southern Mississippi, United States ...... 48

Figure 3.4 Diameter distribution of all Atlantic white-cedar stems within the study site in Southern Mississippi, United States ...... 49

Figure 3.5 Descriptive analyses of Atlantic white-cedar within the study site by crown class and generalized habitat in Southern Mississippi, United States ...... 50

Figure 3.6 Competing overstory and midstory vegetation within the Atlantic white-cedar study site in Southern, Mississippi, United States ...... 51

Figure 4.1 Reproductive and vegetative morphology of Miller’s witch-alder in Baldwin County, Alabama, United States ...... 80

Figure 4.2 Locations of Miller’s witch-alder populations sampled in Alabama and Florida, United States ...... 81

Figure 4.3 Plot diagram for sampling of Miller’s witch-alder habitat ...... 82

Figure 4.4 Stem counts of Miller’s witch-alder populations in Baldwin County (Alabama), Okaloosa County (Florida), and Geneva County (Alabama) ...... 83

Figure 4.5 Habitat suitability model developed for Miller’s witch-alder in the southeastern United States ...... 84

Figure 4.6 Healthy and infected seed capsules of Miller’s witch-alder Baldwin County, Alabama, United States ...... 85 vii

CHAPTER I

INTRODUCTION

Across taxonomic groups, species are currently facing a potential sixth mass extinction event (MacCallum 2015, McCallum 2021, Schachat and Labandeira 2021, Westwood et al.

2021). The United Nations (U.N.) ranks the top five causes of biodiversity loss as; changes in land use, overharvesting, climate change, and invasive species (Diaz et al. 2019). Plants are no different. Human activities are resulting in an increasing number of plant species extinctions around the world (Ellis et al. 2012, Pimm 2020). For example, since European colonization in

North America, there have been 65 known plant extinctions in the United States and Canada

(Knapp et al. 2021). With ever-increasing threats to plant biodiversity, understanding the ecology and natural history of rare and imperiled plants is crucial to ensuring they perpetuate on the landscape. The overall objective of this thesis is to contribute to the ecological understanding and better inform conservation decision making for three rare and imperiled native Southeastern U.S. woody plant species, mountain stewartia (Stewartia ovata (Cav.) Weath.), Atlantic white-cedar

(Chamaecyparis thyoides (L.) B.S.P.), and Miller’s witch-alder (Fothergilla milleri W. D.

Phillips & J. E. Haynes).

To better conserve and protect rare and imperiled plant species, new and innovative methodologies will be needed. One such methodology that has become popular in plant conservation is habitat suitability modeling. Habitat suitability models allow conservationists to focus limited conservation funding and manpower on specific areas with higher habitat

1

suitability to protect a species more efficiently. Areas of high habitat suitability with extant populations can be prioritized for protection. Alternatively, areas with high habitat suitability that do not currently support the species of interest could be prioritized for out-planting to establish new populations. Within this thesis, habitat suitability models were developed for both mountain stewartia and Miller’s witch-alder. These models can therefore be used to inform conservation decision-making for both species.

Disturbance, such as fire or tropical cyclones, are critical forces in many terrestrial systems. Further, many woody plant species require some form of disturbance to regenerate successfully. One common example is serotinous cones on some coniferous tree species. These cones are covered in resin that keeps the cone closed until a fire melts the resin and releases the seed, ensuring the seed is deposited in appropriate growing conditions post- fire (Goubitz et al.

2002). Further, disturbances can open overstory canopies allowing more light to reach the forest floor. Evidence suggests Atlantic white-cedar may require disturbances that allow light to reach the forest floor to regenerate effectively. Fire has been suggested as the main disturbance mechanism by which Atlantic white-cedar regenerates (Korstian 1924). However, fire may not be as prevalent within some populations of Atlantic white-cedar. Other disturbances such as tropical cyclones may be the disturbance needed for regeneration of Atlantic white-cedar in these systems. Through a case study on the Mississippi Gulf Coast, we explore tropical cyclones as a mechanism for regeneration of the Atlantic white-cedar. Understanding the role tropical cyclones play in the regeneration of this imperiled tree species is critical as climate change increases both the number and intensity of tropical cyclones hitting the Gulf Coast.

Plant reproduction occurs one of two ways in most vascular plant species, either sexually or asexually. Sexual plant reproduction occurs through the fusion of male and female gametes.

2

Alternatively, asexual reproduction or vegetative reproduction does not involve gametes and occurs through binary fission. Both reproductive strategies have can be beneficial for a given species in specific situations, however, both strategies can also have drawbacks. For example, asexual reproduction can reduce genetic diversity, while sexual reproduction can be resource intensive. The reproduction strategy utilized by species can have conservation implications that must be understood to properly conserve and manage a species. Anecdotal reports of aphid infestations in the seed of Miller’s witch-alder have recently come to the attention of conservationists. Aphid infestation could severely inhibit the ability of Miller’s witch-alder to reproduce sexually forcing the species to only reproduce vegetatively. This could have sustained negative impacts on the dispersal and genetic diversity within the species. Within this thesis, we assessed the regeneration potential of Miller’s witch-alder. Additionally, given a lack of general ecological information in the primary literature, the microsite habitat requirements of the species were assessed.

Overall, this thesis contributes to the understanding and informs conservation decision- making for three Southeastern U.S. woody plant species. Insights into the regeneration potential of Atlantic white-cedar and Miller’s witch-alder will allow land managers to better understand the ecological requirements of the species and the threats they face. Further, the habitat suitability maps presented within this thesis for mountain stewartia and Miller’s witch-alder can facilitate effective conservation action for the species. Lastly, this work can facilitate additional research into the regeneration potential and development of habitat suitability models for other rare and imperiled plant species to better protect and preserve native flora.

3

Literature Cited

Diaz, S., J. Settle, and E. Brondizio. 2019. Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services (IPBES) Summary for policymakers of the Global Assessment Report on Biodiversity and Ecosystem Services. Intergovernmental Science- Policy Platform on Biodiversity and Ecosystem Services: Bonn, Germany.

Ellis, E.C., E.C. Antill, and H. Kreft. 2012. All is not loss: plant biodiversity in the Anthropocene. PloS One 7(1): e30535.

Goubitz, S., M.J.A. Werger, and G. Ne’eman. 2002. Germination response to fire-related factors of seeds from non-serotinous and serotinous cones. Plant Ecology 169:195-204.

Knapp, W.M, A. Frances, R. Noss, R.F.C. Naczi, A. Weakley, G.D. Gann, B.G. Baldwin, J. Miller, P. McIntyre, B.D. Mishler, G. Moore, R.G. Olmstead. A. Strong, K. Kennedy, B. Heidel, and D. Gluesenkamp. 2021. Vascular plant extinction in the continental United States and Canada. Conservation Biology 35(1):360-368.

Korstian, C.F. 1924. Natural regeneration of southern white cedar. Ecology. 5:188-191.

MacCallum, M.L. 2015. Vertebrate biodiversity losses point to a sixth mass extinction. Biodiversity and Conservation 24:2497-2519.

MacCallum, M. L. 2021. Turtle biodiversity losses suggest coming sixth mass extinction. Biodiversity and Conservation 30:1257-1275.

Pimm, S.L. 2020. What we need to know to prevent a mass extinction of plant species. Plants People Planet 3(1):7-15.

Schachat, S.R. and C.C. Labandeira. 2021. Are heading toward their first mass extinction? Distinguishing turnover from crisis in their fossil record. Annals of the Entomological Society of America 114(2):99-118.

Westwood, M., N. Cavender, A. Meyer, and P. Smith. 2021. Botanic garden solutions to the plant extinction crisis. Plants People Planet 3(1):22-32.

4

CHAPTER II

MODELING HABITAT SUITABILITY FOR STEWARTIA OVATA ACROSS THE

SOUTHEASTERN UNITED STATES

Abstract

Mountain stewartia (Stewartia ovata) is a rare shrub or small tree endemic to the higher elevation regions of Georgia, Tennessee, and Alabama with isolated populations occurring in

Kentucky, North Carolina, South Carolina, Virginia, and Mississippi, USA. The species is often misidentified or overlooked by land managers and conservationists. As a result, mountain stewartia’s habitat and distribution descriptions are limited for restoration and conservation use.

Modeling a species’ habitat suitability has become a critical first step in conserving rare and imperiled plant species. These models allow conservationists to locate previously undocumented populations and prioritize populations and habitats for conservation. This study presents a habitat suitability model for mountain stewartia across its known natural range based on maximum entropy (Maxent) modeling with nine environmental predictor variables and 60 occurrences from herbarium records (n = 22), research-grade iNaturalist observations (n = 25), and other author identified locations (n = 13). The resulting habitat suitability map was classified into bins for spatial analysis. A total of 376,030 hectare (ha) (0.44% of the study area) was designated within the top tier bin with the highest suitable habitat. Further, 133,344 ha (0.16% of the study area) of the top bin was found on publicly owned lands, indicating approximately 35.56% of the highest habitat suitability occurs within public lands. The presented model could allow plant

5

conservationists to prioritize areas for conservation, reintroduction, and may lead to the discovery of previously undocumented populations.

Introduction

Stewartia ovata (Cav.) Weatherby (Theaceae) (Figure 2.1), also referred to as mountain stewartia, summer-flowering dogwood, or mountain camellia is an uncommon, small tree or shrub and grows ca. 300-800 meters (m) in elevation (Granger et al.2018). Mountain stewartia is typically found growing in mixed hardwood and pine stands in proximity to bodies of water

(Granger et al. 2018). Mountain stewartia is predominantly found within the eastern United

States of Alabama, Georgia, North Carolina, and Tennessee. Isolated populations are also known to exist in Kentucky, South Carolina, Mississippi, and Virginia (Kobuski 1951). Mountain

Stewartia populations are often inaccessible due to rugged terrain or river systems, limiting the ability of conservationists to monitor and protect extant populations. Further, a lack of public awareness of the species leads to the species being overlooked by many conservationists and land managers. The lack of public awareness and its overall rarity has led to a knowledge gap regarding the species’ habitat requirements and population stability.

Previous research by Granger et al. (2018) documented little to no sexual reproduction or recruitment in mountain stewartia populations in east Tennessee, raising concerns for the long- term stability of the species. Mountain stewartia is not currently listed as a federally endangered species in the United States. However, the species is listed as critically imperiled in the state of

Mississippi; imperiled in Alabama, South Carolina, and Virginia; and vulnerable in Kentucky,

North Carolina and Georgia. In Tennessee, the conservation status of the species is currently under review (Natureserve Explorer 2020). While extinction is currently unlikely, local extirpation of populations remains possible. Mountain stewartia populations have been reported

6

to be disjunct and isolated from one another (Little 1977). More research is needed to determine the long-term effects, however, the loss of a single disjunct population could reduce the overall genetic diversity within the species and hinder conservation efforts to preserve the species across its native range (Spielman et al. 2004).

Modeling habitat suitability has become a critical first step in the conservation of rare plants (Kumar and Stohlgren 2009, Yang et al. 2013, Remya et al. 2015, West et al. 2016,

Sharma et al. 2018, Ramirez-Reyes et al. 2021). Habitat suitability models, based on known species locations and environmental predictor variables, allow conservationists to identify and protect land that may support a given species (Costa et al. 2010). However, rare species may not appear in traditional botanical surveys and therefore, will have a small number of documented occurrences that are insufficient for use in habitat suitability modeling. Obtaining occurrence records through other means is often necessary for modeling purposes. When data is insufficient, herbarium records and citizen science observations may be used in tandem to increase the number of occurrence records. The increase in the number of observations from herbarium records and citizen science records can better constrain habitat suitability models, producing more accurate and robust models (Chardon et al. 2015, Shaw 2019). One approach to modeling the habitat suitability for a species is to use maximum entropy (Maxent) models (Phillips et al.

2006). Maxent has been shown to perform well relative to other techniques for species with a relatively small sample size (Hernandez et al. 2006, Wisz et al. 2008). Improved habitat suitability models can facilitate research and conservation activities that may insure the long- term sustainability of a species (Hemati et al. 2020).

Other than the description of microsites supporting mountain stewartia by Granger et al.

(2018), little is known regarding the ecological needs of the species. This project aimed to

7

develop a habitat suitability map for mountain stewartia that can be used by conservationists and land managers to preserve mountain stewartia across the landscape. Further, this study contributes to the understanding of the natural history of this relatively little-known species.

Methods

Study Area

Habitat suitability for mountain stewartia was modeled across Alabama, Georgia,

Kentucky, Mississippi, North Carolina, Tennessee, South Carolina, and Virginia (Figure 2.2).

These eight states encompass a total area of 85,786,070 hectares (ha) and contain all known occurrences of mountain stewartia. To summarize the proportion of suitable habitat post- modeling on public land within these states all national parks, national forests, state parks, and state forests were considered (ESRI 2019).

Species Occurrence Data

A total of 104 occurrence points were initially considered with 60 points selected for use in the final model. Occurrences were obtained from herbarium records (n = 36), citizen science records (n = 43), and author identified populations (n = 25). The herbarium records for mountain stewartia were obtained from the Southeast Regional Network of Expertise and Collections

(SERNEC) if the locality of the record was not protected and coordinates were publicly available

(SERNEC 2020) (Figure 2.2). Research-grade iNaturalist occurrence points were also included

(iNaturalist 2020). The iNaturalist application is a citizen science-based program that allows the general public to upload images of species that can then be used as occurrence points by researchers. For an occurrence point to be considered “Research Grade,” two other users must identify the images as the same species as the original author. Additional occurrence points were included from documented populations identified by the authors during previous field visits. To

8

avoid autocorrelation within the occurrence records, we ensured no two occurrences were within the same 4 kilometers squared (km2) raster cell, representing the original pixel size of the climate data. This was accomplished by using a larger subset of occurrences spatially rarified to 4 km using the SDMtoolbox (Brown et al. 2017). Forty occurrence records were excluded due to this spatial rarefication. Additionally, four more occurrences were omitted due to a lack of environmental data at their geographic location. After this process, 60 occurrence records remained for modeling, including herbarium records (n = 22), citizen science records (n = 25), and author identified populations (n = 13). Due to the species’ rarity the authors wish to keep the exact coordinates confidential. Interested parties are welcome to contact the corresponding author for additional information.

Environmental Predictor Data

Nine of the 13 considered environmental layers were selected for use in the final model

(Table 2.1). ArcGIS Pro 2.4.3 was used to prepare all environmental layers. The United States

Geological Survey National Elevation Dataset one arc-second with 30 m resolution was used for elevation (U.S. Geologic Survey 2017). Heat load was calculated according to McCune and

Keon (2002) utilizing the heat load tool within the ArcGIS Geomorphometry & Gradient Metrics toolbox (Brown et al. 2017). Heat load values close to zero represent the cooler, northeast aspects, and values closer to one represent the warmer, southwest aspects. The Euclidean distance, or straight-line distance, from a stream or body of water was used within the model.

The USA Detailed Streams layer was rasterized and input into the Euclidean distance function

(Gary et al. 2009). This resulting raster indicated the number of raster cells a specific cell is away from a known stream or body of water. Climatic data from the PRISM Climate Group representing the 30-year normal annual values (1981 - 2010) were used at 4 km resolution for

9

three climatic environmental layers, including mean annual precipitation, mean maximum annual temperature, and mean minimum annual vapor pressure deficit (PRISM Climate Group 2020).

Soil pH and percent soil sand content were downloaded from the SSURGO database at 30 m resolution for use within the model (Soil Survey Staff 2019). Using the CONUS 2016 national landcover dataset, the proportion of evergreen forest cover within a 100 meter (m) radius was calculated for each cell (Yang et al. 2019).

All environmental layers were cropped to the dimensions of the study area and spatial resolution was adjusted to match a 100x100 m pixel size using the nearest neighbor resampling method. All datasets were projected to the North America Albers Equal Area Conic projection.

All uncorrelated environmental variables were used within the model (|r| < 0.7, Dormann et al.

2013). The correlation and summary stats tool within the SDMtoolbox was used to calculate correlation among dataset (Brown et al. 2017). Four additional variables were considered but dropped from the model. Yearly maximum vapor pressure deficit and yearly minimum temperature were both highly correlated with yearly maximum temperature and were therefore not necessary in the final model. Soil organic material was considered as an environmental predictor variable but was dropped from the model due to a lack of coverage. Topographic wetness index as calculated by Beven and Kirby (1979) was also considered but was ultimately not used as distance from a water body was a better predictor of mountain stewartia habitat suitability.

Maxent Model

Maxent version 3.4.0 was used to develop the presented model (Phillips et al. 2006).

These models utilize occurrence records in the form of geographic coordinates and environmental grid layers to estimate a species’ habitat suitability across a defined region.

10

Maxent utilizes presence-only data as opposed to presence-absence data (Phillips et al. 2006,

Phillips and Dudík 2008). Presence-only models allow for the use of opportunistic surveys that do not record species absences. Maxent allows users to change the regularization multiplier within the software. The regularization multiplier protects against overfitting by penalizing the model for each additional term and for higher weights given to those terms (Phillips et al. 2006,

Anderson and Gonzalez 2011, Radosavljevis and Anderson 2014). The default regularization multiplier of one within the Maxent software has been shown to overfit models and a higher multiplier is needed in some cases (Radosavljevic and Anderson 2014, Galante et al. 2018). A range of regularization multipliers were utilized as suggested by Merow at al. (2013) and

Radosavljevis and Anderson (2014) with a 60-fold cross-validation test to maximize the area under the receiving operator curve (AUC). The AUC is a measure used to evaluate a model’s goodness-of-fit, and a model with an AUC value greater than 0.70 is generally considered useful

(Swets 1988). A regularization multiplier of one was selected for use in the final model, as it optimized AUC. Maxent utilizes background data (or pseudo-absences) to contrast known occurrences to the surrounding landscape where there are no known occurrences (Merow at al.

2013, Radosavljevis and Anderson 2014). By default, Maxent assumes a uniform prior for background data, in other words, all pixels within the study area have the same probability of being selected for pseudo-absences (Merow et al. 2013). For large study areas, this has been shown to artificially inflate AUC values (VanDerWal et al. 2009, Merow et al. 2013). To avoid artificially inflating the AUC for this study, background data were confined to only hydrologic unit code (HUC) 8 level watersheds that contain known mountain stewartia populations (Gary et al. 2009). The model was then projected out for all states documented to contain populations of mountain stewartia. A uniform prior was assumed within the watersheds containing mountain

11

stewartia (Merow et al. 2013). In total 5,000 background points were used in the modeling process. All other model parameters were retained at their default values. The jackknife procedure within the Maxent software was used to determine the importance of the input environmental data layers.

The model was run six additional times to access the effect of various environmental variable groups within the model. Environmental variables were classified into three groups: soil variables, climatic variables, and physical variables. The soil variable group included soil pH and percent soil sand content. The climatic variable group included precipitation, maximum temperature, and minimum vapor pressure deficit. The physical variable group included distance to water, proportion of evergreen cover, elevation, and heat load. All variable groups were run alone and in all possible combinations.

Post-modeling, Maxent provides a habitat suitability map of the study area in a raster dataset format. Each pixel within the map has a calculated habitat suitability from zero (low suitability) to one (high suitability). The pixels were then binned into discrete classes for each tenth of estimated habitat suitability, where the 0.9-1.0 bin was considered the highest habitat suitability.

Results

The model was determined to be useful for explaining the observed distribution of mountain stewartia (AUC = 0.80). A total 376,030 ha (0.44% of the study area) was within the highest habitat suitability bin. The physiographic region with the most area within the highest habitat suitability bin was the Blue Ridge Mountains (Bailey 2016) (Figure 2.2). The physiographic regions, as defined by Bailey (2016), with the second and third most area within the highest habitat suitability bin were the Northern Cumberland Plateau and Southern

12

Cumberland Plateau, respectively (Table 2.2). The hydrological unit code (HUC) 10 watershed with the most area within the highest habitat suitability bin was the Emory River watershed (TN) with 7,148 ha (Table 2.3). The watersheds with the second and third most area within the highest habitat suitability bin were the Straight Creek-Upper Little River (GA/AL) watershed and New

River (TN) watershed with 6,897 ha and 6,118 ha, respectively. Approximately 35.56% (133,344 ha) of this highest bin was identified as public-owned land. Based on percent contribution to the model, distance to water (44.4%), percent soil sand content (26.3%), mean maximum temperature (12.5%), and mean minimum vapor pressure deficit (5.9%) were found to be the most important predictor variables for the model (Table 2.1).

The test AUC did not vary greatly as the regularization multiplier was adjusted. It peaked at a regularization multiplier of one and dropped from 0.80 to 0.79 when the regularization multiplier was increased to two. The test AUC was the lowest at a regularization multiplier of ten at 0.74. The test AUC did vary when variable groups were modeled alone and in all possible combinations with one another. Physical variables alone produced the lowest AUC of 0.70, while soil variable in combination with climatic variables produced an AUC of 0.78. When climatic variables were modeled alone, they produced an AUC of 0.70. Soil variables modeled alone produced an AUC of 0.71. Soil and physical variables produced an AUC of 0.77. Lastly, climatic and physical variables produced an AUC of 0.75.

Discussion

Known populations of mountain stewartia grow primarily on the Northern Cumberland

Plateau and Southern Cumberland Plateau, in Kentucky, Tennessee, and Alabama and the Blue

Ridge Mountains of Tennessee, North Carolina, and Georgia. These regions were modeled with the highest habitat suitability (Figure 2.2). Based upon the model and previous work by Granger

13

et al. (2018), the Northern Cumberland Plateau, Southern Cumberland Plateau and Blue Ridge

Mountains may provide the most optimal temperature ranges, soil conditions, and moisture levels necessary for the establishment and growth of mountain stewartia (Figure 2.2). More precisely, specific watersheds within these physiographic sub-regions, such as the Emory watershed (TN), Straight Creek-Upper Little River watershed (AL/GA), and New River watershed (TN), likely provide the same optimal abiotic requirements for mountain stewartia establishment and growth. These hotspots, at both the physiographic sub-region and watershed levels, can thus be prioritized for the conservation and restoration of the species.

Percent sand content and Euclidean distance to water were the most important variables in the habitat suitability model, matching findings from previous research. As described by

Granger et al. (2018), mountain stewartia prefers highly acidic, permeable, well-drained, cobbly loam soils. Percent soil sand content contributed the second most of the nine environmental variables (26.3%). These findings support the assertion by Granger et. al (2018) that mountain stewartia prefers specific soil qualities. However, the overall contribution of soil pH to the model was found to be relatively low (5.5%). Therefore, this result was less consistent with the findings of Granger et al. (2018). Lastly, mountain stewartia is known to grow near streams and creeks

(Granger et al. 2018). Euclidean distance to water was the highest contributor to the overall model (44.4%), indicating mountain stewartia prefers microclimates associated with creeks and streams.

Field-based assessments of mountain stewartia habitat have indicated the species is associated with large diameter eastern white pine (Pinus strobus L.) and eastern hemlock snags

(Tsuga canadensis (L.) Carrière) (Granger et al. 2018). Proportion of evergreen species within the overstory did not highly contribute to the overall model (0.2%). However, this may not

14

reflect the true ecological relationship between evergreen species and mountain stewartia. Litter and organic matter from many evergreen species, especially conifers, can lower the cation concentration of soil leading to the acidification of forest soils (Brandtberg et al. 2000). Initially, both percent organic matter in the soil and pH were used in the model along with the proportion of evergreen species in the overstory to represent this complex relationship between evergreen species, soil pH, organic matter, and mountain stewartia. However, percent organic matter in the soil was not included in the final model due to a lack of data coverage across the study area.

Also, data collection protocols for organic material typically do not include leaf litter and woody debris within the O horizon. This may not properly assess the relationship between mountain stewartia and leaf litter or woody debris in this horizon. Soil pH was included as it contributed to the overall model by 5.5% and was not highly correlated with proportion of evergreen species within the overstory (|r| = 0.12). The complex relationship between evergreen species composition, soil pH, organic matter, and mountain stewartia was difficult to incorporate into the model due to a lack of available landscape-scale environmental covariates across the study area.

Additional field work and advances in landscape-level data are needed to understand this complex relationship and to be able to more effectively incorporate them into these habitat suitability models.

Isolated populations of mountain stewartia are known to occur in coastal Virginia (Little

1977). These are the only documented populations known to occur along coastal communities.

GPS coordinates for these populations were unavailable for use in the presented model.

However, this model did identify suitable habitats within the coastal regions of North Carolina and Virginia (Figure 2.2). These findings support the applicability of this model beyond just the occurrence points used.

15

Public lands are critical for plant conservation efforts (Havens et al. 2014). Due to a perceived contradiction between economic productivity and biodiversity conservation, it may be difficult to motivate private landowners to manage their land for conservation (Doremus 2003).

Identifying appropriate habitat for plant species on public lands better ensures the long-term success of conservation projects. Of the 376,030 ha classified within the top bin of the model,

133,344 ha (35.5%) corresponded to public land with some level of protection. The presented model allows for the prioritization of land that is both ecologically appropriate and under the jurisdiction of public land management agencies.

The use of herbaria and citizen science records are regarded for their usefulness in supporting ecological research (Fosberg 1946, Sauer 1953, Loiselle et al. 2008, Silvertown 2009,

Miller-Rushing et al. 2012, Greve et al. 2016). Lloyd et al. (2020) suggested that in some situations, citizen science records can be highly valued for the conservation of rare or threatened species. Recently, there have been initiatives to protect and digitize herbarium collections and develop online tools (e.g., iNaturalist and eBird) (Barkworth and Murrell 2012, Thiers et al.

2016). The use of online tools can provide both a platform to engage the at-large community and standardized datasets for research (Conrad and Hilchey 2011, Theobald et al. 2015). Moving forward, models built on older records may need to be interpreted cautiously as they may no longer represent the suitable habitat for a given species with shifting climate conditions. The model presented here adds further support to the growing body of evidence for the need of herbariums and citizen science records to support basic ecological research. Without these records, the presented model would not be as useful due to a lack of occurrence data. The ecological understanding gained from these collections should allow conservationists to improve decision-making and reduce costs (McKinley et al. 2017).

16

Conclusions

When used in conjunction with habitat suitability models, herbarium and citizen science records may identify under-sampled locations for a given species and lead to the discovery of unrecorded populations. They may also be used to identity suitable areas, currently void of the desired plant species, that could be prioritized for introduction or reintroduction efforts.

Conversely, habitat suitability models may be used in the prioritization of known populations for conservation and restoration efforts.

Acknowledgements

We wish to thank David S. Buckley, Jack Johnston, Ron Miller, and Sara Johnson for their review and contributions to this project. This work is the combination of the Forest and

Wildlife Research Center and the Mississippi Agriculture and Forest Experiment Station,

Mississippi State University and this material is based upon work that is supported by the U.S.

Department of Agriculture, National Institute for Food and Agriculture, McIntire-Stennis project under accession number 1019117. Any opinions, findings, conclusions, or recommendations are those of the authors and do not necessarily reflect the views of the U.S. Department of

Agriculture.

17

Tables

Table 2.1 Contribution of environmental variables to mountain stewartia habitat suitability model in the Southeastern United States.

Environmental Variables Source Contribution % Mean (SD) Unit

Euclidean Distance to Water U.S. Geologic Survey 44.1 432.21 (608.30) Meters

Percent Sand within the Soil U.S.N.R.C.S. 26.3 5.01(1.71) Percent

Mean Yearly Max. Temp. PRISM Climate Group 12.5 20.35 (1.26) Celsius

Mean Yearly Min. VPD PRISM Climate Group 5.9 0.62(0.17) Kilopascal

Soil pH U.S.N.R.C.S. 5.5 5.14 (0.34) Unitless

Mean Yearly Precip. PRISM Climate Group 3.3 1,454.96 (135.92) Millimeter

Elevation U.S. Geological Survey 1.0 401.93 (219.47) Meters

Heat Load U.S. Geological Survey 0.9 0.85 (0.02) Unitless

Proportion of Evergreen Species National Landcover Database 0.2 8.29 (17.01) Percent

Overall percent contribution, mean, and standard deviation for the environmental variables used in the maxent habitat suitability modeling of (Stewartia ovata (Cav.) (Theaceae)). U.S.N.R.C.S. represents the U.S. Natural Resource Conservation Service.

18

Table 2.2 Physiographic sub-ecoregions with the greatest total area identified within the top bin of habitat suitability model for mountain stewartia in the southeastern United States

Physiographic Sub-Ecoregion Area (ha)

Blue Ridge Mountains 128,960

Northern Cumberland Plateau 82,972

Southern Cumberland Plateau 51,203

Coastal Plains, Middle 23,406

Interior Low Plateau, Highland Rim 21,970

Upper Gulf Coastal Plain 19,283

Central Ridge and Valley 17,891

Southern Cumberland Mountains 15,116

Southern Appalachian Piedmont 7,105

Southern Ridge and Valley 5,492

The ten physiographic sub-ecoregions with the greatest total area identified within the top bin of habitat suitability model for mountain stewartia (Stewartia ovata (Cav.) (Theaceae)). Physiographic sub-ecoregions are defined by Bailey's Ecoregions and Subregions of the United States, Puerto Rico, and the U.S. Virgin Islands.

19

Table 2.3 Watersheds with the greatest total area identified within the top bin of habitat suitability model for mountain stewartia in the Southeastern United States

Total area

Area (ha) within top bin of (ha) of

Watershed habitat suitability watershed

Emory River (TN)* 7,148 136,964

Straight Creek-Upper Little River (AL/GA)* 6,897 40,126

New River (TN) 6,118 158,171

Guntersville Lake-Town Creek (AL) 5,846 82,310

Mud Creek-Tennessee River (AL)* 5,719 114,035

Ocoee River (GA/NC/TN) 5,483 157,612

Tallulah River (GA/NC) 5,410 73,006

Upper Bear Creek (AL/MS)* 5,168 127,375

Headwaters Little Tennessee River (GA/NC) 5,141 76,912

Nottely River-Nottely Lake (GA) 4,771 82,597

Toccoa River-Blue Ridge Lake (GA) 4,625 89,113

Headwaters Pigeon River (NC) 4,387 65,705

Chattooga River (GA/NC/SC)* 4,289 107,770

Valley River (NC) 4,072 45,511

Davidson River-French Broad River (NC/SC) 3,936 65,203

The 15 hydrological unit code (HUC) 10 watersheds, as defined by the U.S. Geological Survey, with the greatest total area identified within the top bin of habitat suitability for mountain stewartia (Stewartia ovata (Cav.) (Theaceae)).

20

Figures

Figure 2.1 Flower and form of mountain stewartia in Southern Appalachia, United States

A) Typical vegetative and reproductive morphology of mountain stewartia (Stewartia ovata (Cav.) (Theaceae)). (B) Typical habitat and growth form of mountain stewartia. Photos taken by Joshua J. Granger.

21

Figure 2.2 Habitat suitability model developed for mountain stewartia in the Southeastern United States

The habitat suitability model developed within Maxent for mountain stewartia (Stewartia ovata (Cav.) (Theaceae)) from known occurrence points (n = 60) and environmental variables (distance to water, percent sand within the soil, 30-year average yearly maximum temperature, 30-year average yearly minimum vapor pressure deficit, 30-year average yearly precipitation, soil pH, elevation, heat load, proportion of evergreen species within the overstory). The study area is indicated by the grey shading. The model had an overall AUC of 0.80.

22

Literature Cited

Anderson, R.P., and I. Gonzalez, Jr. 2011. Species-specific tuning increases robustness to sampling bias in models of species distributions: an implementation with maxent. Ecological Modeling 222:2796-2811.

Bailey, R.G. 2016. Bailey’s ecoregions and subregions of the United States, Puerto Rico, and U.S. Virgin Islands. U.S. Forest Service, United States Department of Agriculture, Washington, D.C. < https://www.fs.usda.gov/rmrs/datasets/baileys-ecoregions-and- subregions-united-states-puerto-rico-and-us-virgin-islands> Accessed August 22, 2021.

Barkworth, M.E. and Z.E. Murrell. 2012. The US virtual herbarium: Working with individual herbaria to build a national resource. Zookeys 209:55-73.

Beven, K. and M. Kirby.1979. A physically based, variable contributing area model of basic hydrology. Hydrological Sciences Bulletin 24(1):43-69.

Brandtberg, P.O., H. Lundkvist, and J. Bengtsson. 2000. Changes in forest-floor chemistry caused by a birch admixture in Norway spruce stands. Forest Ecology and Management 130:253–64.

Brown J.L., J.R. Bennett, and C.M. French. 2017. SDMtoolbox 2.0: the next generation Python- based GIS toolkit for landscape genetic, biogeographic and species distribution model analyses. Peerj 5:e4095.

Chardon, N.I., W.K. Cornwell, L.E. Flint, A.L. Flint, and D.D. Ackerly. 2015. Topographic, latitudinal and climatic distribution of Pinus coulteri: geographic range limits are not at the edge of the climate envelope. Ecography 38(6):590-601.

Conrad, C.C., and K.G. Hilchey. 2011. A review of citizen science and community-based environmental monitoring: issues and opportunities. Environmental Monitoring and Assessment 176:273-291.

Costa, G.C., C. Nogueira, R.B. Machado, and G.R. Colli. 2010. Sampling bias and the use of ecological niche modeling in conservation planning: a field evaluation in a biodiversity hotspot. Biodiversity and Conservation 19:883-899.

Doremus, H. 2003. A policy portfolio approach to biodiversity protection on private lands. Environmental Science and Policy 6(3):217-232.

Dormann, C.F., J. Elith, S. Bacher, C. Buchmann, G. Carl, G. Carré, J.R.G. Marquéz, B. Gruber, B. Lafourcade, P. J. Leitão, T. Münkemüller, C. McClean, P.E. Osborne, B. Reineking, B. Schröder, A.K. Skidmore, D. Zurell, and S. Lautenbach. 2013. Collinearity: a review of methods to deal with it and a simulation study evaluating their performance. Ecography 36:27-46.

23

ESRI. 2019. USA Parks. Accessed 26 April 2020

Fosberg, F.R., 1946. The Herbarium. Scientific Monthly. 63(6):429-434.

Galanta, P.J. B. Alade, R. Muscarella, S.A. Jansa, S.M. Goodman, and R.P. Anderson. 2018. The challenge of modelling niches and distributions for data-poor species: a comprehensive approach to model complexity. Ecography 41:726-736.

Gary, R.H., Z.D. Wilson, C.M. Archuleta, F.E. Thompson, and J. Vrabel. 2009. Production of a 1:1,000,000-scale hydrography dataset for the United States – feature selection, simplification, and refinement. U.S. Geological Survey Scientific Investigations Report. 2009-5202. 22.

Granger, J.J., D.S. Buckley, and J.M. Zobel. 2018. Microsites supporting endemic populations of mountain stewartia (Stewartia ovata) in East Tennessee. Castanea 83:140-151.

Greve, M., A.M. Lykke, C.W. Fagg, R.E. Gereau, G.P. Lewis, R. Marchant, A.R. Marshall, J. Ndayishimiye, J. Bogaert, and J.C. Svenning. 2016. Realizing the potential of herbarium records for conservation biology. South African Journal of Botany 105:317-323.

Havens K., A.T. Kramer, and E.O. Guerrant Jr. 2014. Getting plant conservation right (or not): the case of the United States. International Journal of Plant Science 175(1):3-10.

Hemati, T., S. Pourebrahim, M. Monavari, and A. Baghvand. 2020. Species-specific nature conservation prioritization (a combination of maxent, co$sting nature, and dinamica ego modeling approaches). Ecological Modeling 429.

Hernandez, P.A., C.H. Graham, L.L. Master, and D.L. Albert. 2006. The effect of sample size and species characteristics on performance of different species distribution modeling methods. Ecography 29(5):773-785. iNaturalist. 2020. Mountain Camellia. (https://www.iNaturalist.org/taxa/169392-Stewartia- ovata, 26 April 2020)

Kobuski, C.E. 1951. Studies in the Theaceae, XXI the species of Theaceae indigenous to the United States. Journal of the Arnold Arboretum 32:123–138.

Kumar, S. and T.J. Stohlgren. 2009. Maxent modeling for predicting suitable habitat for threatened and endangered tree Canacomyrica monticola in New Caledonia, Journal of Ecology and Natural Environment 1(4):94-98.

Little, E.L. Jr. 1977. Atlas of United States Trees. Volume 4. Minor Eastern Hardwoods. U.S. Department of Agriculture, Forest Service, Washington, D.C.

24

Lloyd, T.J., R.A. Fuller, J.L. Oliver, A.I. Tulloch, M. Barnes, and R. Stevens. 2020. Estimating the spatial coverage of citizen science for monitoring threatened species. Global Ecology and Conservation 23:e01048.

Loiselle B.A., P.M. Jørgensen, T. Consiglio, I. Jiménez, J.G. Blake, L.G. Lohmann, and O.M. Montiel. 2008. Predicting species destributions from herbarium collections: does climate bias in collection sampling influence model outcomes? Journal of Biogeography 35(1):105-116.

McCune, B. and D. Keon. 2002. Equation for potential annual direct incident radiation and heat load. Journal of Vegetation Science 13:603-606.

McKinley D.C., A.J. Miller-Rushing, H.L. Ballard, R. Bonney, H. Brown, S.C. Cook-Patton, D.M. Evans, R.A. French, J.K. Parris, T.B. Phillips, S.F. Ryan, L.A. Shanley, J.L. Shirk, K.F. Stepenuck, J.K. Weltzin, A. Wiggins, O.D. Boyle, R.D. Briggs, S.F. Chapin III, D.A. Hewitt, P.W. Preuss, and M.A. Soukup. 2017. Citizen science can improve conservation science, natural resource management, and environmental protection. Biological Conservation 208:15-28.

Merow, C., M.J. Smith, and J.A. Silander Jr. 2013. A practical guide to maxent for modeling species’ distributions: what it does, and why inputs and settings matter. Ecography 36:1058-1069.

Miller-Rushing A., R. Primack, and R. Bonney. 2012. The history of public participation in ecological research. Frontiers in Ecology and the Enviroment 10:285-290.

NatureServe Explorer. 2020. An Online Encyclopedia of Life: Mountain Camellia. (https://explorer.natureserve.org/Taxon/ELEMENTGLOBAL.2.148162/Stewartia_ovata, 26 April 2020)

Phillips, S.J., R.P. Anderson, and R.E. Schapire. 2006. Maximum entropy modeling species geographic distributions. Ecological Modeling 190:231-259.

Phillips S.J. and M. Dudík. 2008. Modeling of species distributions with MaxEnt: new extensions and a comprehensive evaluation. Ecography 31:161–175

PRISM Climate Group. 2020. 30-Year Normal. Oregon State University. Accessed, 20 March 2020.

Radosavljevic, A., and R.P. Anderson. 2014. Making better maxent models of species distributions: complexity, overfitting, and evaluation. Journal of Biogeography 41:629- 643.

25

Ramirez-Reyes, C., G. Street, F. Vilella, D.T. Jones-Farrand, M.S. Wiggers, and K.O. Evans. 2021. Ensemble species distribution model identifies survey opportunities for at-risk bearded beaksedge (Rhynchospora crinipes) in the southeastern United States. Natural Areas Journal 41(1):55-63.

Remya, K., A. Ramachandran, and S. Jayakumar. 2015. Predicting the current and future suitable habitat distribution of Myristica dactyloides Gaertn. using MaxEnt model in the Eastern Ghats, India. Ecological Engineering 82:184-188. Sauer, J.D., 1953. Herbarium Specimens as Records of Genetic Research. American Naturalist 87(834):155–156.

SERNEC. 2020. Data Portal Stewartia ovata. (http://sernecportal.org/portal/, 6 February 2020)

Sharma, S., K. Arunachalam, D. Bhavsar, and R. Kala. 2018. Modeling habitat suitability of Perilla frutescens with maxent in Uttarakhand-A conservation approach. Journal of Applied Research on Medicinal and Aromatic Plants 10:99–105.

Shaw, J. 2019. A reevaluation of Tennessee’s invasive plant species using SERNEC and maximum entropy species distribution models. M.S. Thesis, University of Tennessee at Chattanooga, Chattanooga, Tennessee.

Silvertown, J. 2009. A dawn for citizen science. Trends in ecology and evolution 24:467-471.

Soil Survey Staff. 2019. Soil Survey Geographic (SSURGO) Database for the United States. Natural Resources Conservation Service, United States Department of Agriculture, Washington, D.C.

Spielman, D., B.W. Brook, and R. Frankham. 2004. Most species are not driven to extinction before genetic factors impact them. Proceedings of the National Academy of the Sciences 101(42):15261–15264.

Swets, J.A. 1988. Measuring the Accuracy of Diagnostic Systems. Science 240(4857):1285- 1293.

Theobald E.J., A.K. Ettinger, H.K. Burgess, L.B. DeBey, N.R. Schimidt, H.E. Froehlich, C. Wagner, J. Hillerislambers, J. Tewksbury, M.A. Harsch, and J.K. Parrish. 2015. Global change and local solutions: tapping the unrealized potential of citizen science for biodiversity research. Biological Conservation 181:236-244.

Thiers, B.M., M.C. Tulig, and K.A. Watson. 2016. Digitization of The New York Botanical Garden Herbarium. Brittonia 68(3):324-333.

26

U.S. Geologic Survey. 2017. 1 Arc-second Digital Elevation Models (DEMs) - USGS National Map 3DEP Downloadable Data Collection. United States Geologic Survey, United States Department of the Interior, Washington, D.C. https://www.sciencebase.gov/catalog/item/4f70aa71e4b058caae3f8de1 Accessed 22 August 2021

VanDerWal, J., L.P. Shoo, C. Graham, and S.E. Williams. 2009. Selecting pseudo-absence data for presence-only distribution modeling: How far should you stray from what you know. Ecological Modeling 220:589-594.

West, A.M., S. Kumar, C.S. Brown, T.J. Stohlgren, and J. Bromberg. 2016. Field validation of an invasive species maxent model. Ecological Informatics 36:126-134.

Wisz, M.S., R.J. Hijmans, J. Li, A.T. Peterson, C.H. Graham, A. Guisan, and NCEAS Predicting Species Distributions Working Group. 2008. Effects of sample size on the performance of species distribution models. Diversity and Distributions 14(5):763-773.

Yang, L., S. Jin, P. Danielson, C. Homer, L. Gass, S.M. Bender, A. Case, C. Costello, J. Dewitz, J. Fry, M. Funk, B. Granneman, G.C. Liknes, M. Rigge, and G. Xian. 2019. A new generation of the United States national landcover database: requirements, research priorities, design, and implementation strategies. Journal of Photogrammetry and Remote Sensing 146:108-123.

Yang, X.Q., S.P.S. Kushwaha, S. Saran, J. Xu, and P.S. Roy. 2013. Maxent modeling for predicting the potential distribution of medicinal plant, Justicia adhatoda L. in Lesser Himalayan foothills. Ecological Engineering 51:83.

27

CHAPTER III

ATLANTIC WHITE-CEDAR (CHAMAECYPARIS THYOIDES) RESPONSE POST-

HURRICANE DISTURBANCE

Abstract

Atlantic white-cedar (Chamaecyparis thyoides) is considered critically imperiled across much of its range. This tree species occurs in freshwater swamps, rarely further inland than 160 kilometers (km) from the Atlantic or Gulf Coasts. This proximity to the coast makes it vulnerable to windthrow disturbances from tropical cyclones. As the frequency and intensity of tropical storm systems increase, understanding the regeneration of Atlantic white-cedar will be critical to its conservation. This study evaluated the regeneration of Atlantic white-cedar 14-years following Hurricane Katrina. All Atlantic white-cedar ≥ 2.5 centimeters (cm) at breast height were inventoried within a ~9 hectares (ha) study area located within Grand Bay National

Wildlife Refuge in Jackson County, Mississippi, USA. Following Hurricane Katrina, the number of Atlantic white-cedar stems increased by 191%. Yet, this may only represent a net increase of

78 stems due to intraspecies competitive exclusion and the encroachment of exotics. Silvicultural efforts, such as hardwood removal, may become necessary to ensure the success of Atlantic white-cedar across the site due to competition. This long-term study provides the first analysis of

Atlantic white-cedar regeneration post-tropical cyclone disturbance within the Gulf Coast and allows conservationists a better understanding of the effects of tropical cyclone disturbance on this species.

28

Introduction

Tropical cyclones are increasing in both number and intensity due to shifts in the climate

(Webster et al. 2005, Elsner and Jagger 2010, Patricola and Wehner 2018, Camargo and Wing

2021, Wang and Toumi 2021). Tropical cyclones are a historic disturbance on coastal plant communities, however, increases in tropical cyclone frequency and intensity are having sustained negative effects on many coastal forested communities, including changes in stand mortality and regeneration patterns (Lugo and Scatena 1996, Lugo 2000, Dale et al. 2001). Little is currently known of how an increase in the frequency and intensity of tropical storms will affect the ecology and regeneration of coastal forested communities or the species found within them. Thus, a better understanding of coastal plant community ecology and their regeneration post tropical cyclone disturbance will be critical to conserving them into the future.

Disturbance, particularly tropical cyclones, play a critical role in the ecology of coastal plant communities (Xi 2015, Ibañez et al. 2019, Lin et al. 2020). In forested systems, tropical cyclones alter the trajectory of forest regeneration (Everham and Brokaw 1996). In many cases, tropical cyclones can alter the physical structure of forest. For example, forested communities with higher frequencies of tropical cyclones tend to have higher stem densities and basal areas

(Ibañez et al. 2019). Tropical cyclones also induce overstory tree mortality, thus opening canopy gaps for shade intolerant species to regenerate (Comita et al. 2009). Lastly, invasive species, such as Chinese tallow (Triadica sebifera (L.) Small), have been shown to exploit the additional light provided by tropical cyclone disturbance (Henkal at al. 2016).

Atlantic white-cedar (Chamaecyparis thyoides (L.) B.S.P.) typically grows in coastal freshwater swamps and bogs with high water tables, resulting in shallow rooting due to the low oxygen environment (Korstian and Brush 1931). The shallow roots of Atlantic white-cedar make

29

it highly vulnerable to windthrow from tropical cyclones. Further, Atlantic white-cedar is primarily found growing in a band 160 kilometers (km) wide along the Atlantic and Gulf Coast, placing many populations directly in the path of tropical cyclones as they make landfall. The proximity to the coast and high potential for windthrow from tropical storms makes Atlantic white-cedar an ideal study species to understand the role of tropical cyclones in the regeneration and ecology of forested coastal plant communities.

Atlantic white-cedar primarily occurs along the Atlantic Coast from Maine to Florida, and in isolated populations along the Gulf of Mexico coastal regions of Florida, Alabama,

Georgia, and Mississippi (Korstian and Brush 1931, Little 1979, Hardin et al. 2000). In

Mississippi, the species is considered imperiled and at risk of extirpation by extreme weather events, exploitation, altered disturbance regimes, and shifts in land use (Natureserve Explorer

2020). Historically, the species was logged heavily, which resulted in population declines and the elimination of natural stands across much of its range (Ward 1989). Changes in hydrologic processes, such as wetland draining for agriculture, and land use are the primary risk factors currently impacting the species (Ehrenfeld and Schneider 1991).

Gulf Coast populations of Atlantic white-cedar have been suggested to be a distinct species, Chamecyparis henryae (H.L. Li) (Li 1962), but many consider it a variant (Korstian and

Brush 1931, Little 1966, Ward and Clewell 1989). Li (1962) reported morphological differences in populations of the Atlantic Coast and Gulf Coast as the basis of the new species. According to

Li (1962), Gulf Coast populations exhibit smoother bark with a distinctive twist in the trunk of mature individuals. The foliage of the Gulf Coast population is reportedly rounder and lighter in color and lacking two distinct white bands on the lower surface (Li 1962). Gulf Coast populations have been reported to have cones that are slightly larger than that of their Atlantic

30

counterparts (Li 1962). Additionally, the ecology of these Gulf Coast and Atlantic Coast populations have been known to differ. On the Atlantic coast, Atlantic white-cedar is generally observed in pure, dense stands (Hardin et al. 2000, Mylecraine and Zimmermann 2000), while

Gulf Coast populations tend to occur in mixed stands with bald-cypress (Taxodium distichum

(L.) Rich.), gum (Nyssa spp.) and ash species (Fraxinus spp.) (Eleuterius and Jones 1972,

McCoy and Keeland 2006, Keeland and McCoy 2007). The morphological and ecological differences of Atlantic and Gulf Coast populations of Atlantic white-cedar could result in divergent best management practices for the two populations.

It is commonly believed that fire is the dominant disturbance type facilitating Atlantic white-cedar regeneration (Korstian 1924, Laderman 1989). Though damaging to mature trees, fire has been shown to facilitate Atlantic white-cedar regeneration by opening the canopy and providing suitable conditions for seed germination (Korstian 1924). However, there is little evidence of fire in wet Atlantic white-cedar stands along the Gulf Coast, USA. Yet, there is evidence of frequent wind disturbance through tropical storm systems. For example, Doyle

(2009) reports that from 1851 to 2006 Category One hurricanes occurred once every eight years, and Category Two hurricanes occurred once every 16 years in southeast Mississippi. While

Nowacki and Abrams (2008) report the woody wetland cover type in eastern United States would rarely burn, averaging over 200 years between fires. Thus, the difference in return intervals of fire and hurricane disturbance could point towards tropical cyclones, rather than fire, as the primary factor affecting Atlantic white-cedar regeneration on the Gulf Coast. One such storm in recent history that could have facilitated Atlantic white-cedar regeneration on the Gulf

Coast is Hurricane Katrina.

31

In 2005, Hurricane Katrina devastated human and ecological communities on a scale rarely seen. Following Hurricane Katrina, McCoy and Keeland (2006) reported on the destruction, but not the regeneration, of a remnant Atlantic white-cedar population near the

Escatawpa River in southeastern Mississippi. Only one study has addressed the regeneration of

Atlantic white-cedar after a large-scale tropical cyclone disturbance event. Following Hurricane

Isabel in September of 2003, Laing et al. (2011) analyzed Hurricane Isabel’s impact on Atlantic white-cedar within the Great Dismal Swamp in North Carolina and Virginia and concluded active management, such as herbicide or prescribed fire, was necessary for the population to regenerate after a large-scale wind disturbance. This study provides additional data as a complement to the results of Laing et al. (2011) by providing insights into how Gulf Coast populations recover after a tropical cyclone disturbance.

The objective of this study was to evaluate the long-term impacts of tropical cyclones on the recovery and regeneration of an Atlantic white-cedar population on the U.S. Gulf Coast. As

Gulf Coast tropical storm systems continue to increase, understanding the role tropical cyclones have on Atlantic white-cedar regeneration potential will be key to ensuring the sustainability of this already imperiled species.

Methods

2006 Post- Katrina Data Source

In 2006, McCoy and Keeland published the results of a survey of an Atlantic white-cedar stand that was directly hit by Hurricane Katerina the year before. They completed a 100% inventory of Atlantic white-cedar within the stand as well as a survey of the associated woody plant community. Using the historical data presented in McCoy and Keeland (2006) as a

32

baseline, we assessed the regeneration of Atlantic white-cedar and changes in the community assemblage for the same study site within Grand Bay National Wildlife Refuge.

Study Site

The study site is in the southeast corner of Jackson County, Mississippi, USA, at approximately 30°27'44.0"N and 88°26'37.4"W. The site is bordered on the north by the

Escatawpa River, on the west by a man-made canal known as Paper Mill Canal, on the south by

Interstate 10, and on the west by a managed longleaf pine (Pinus palustris Mill.) and slash pine

(Pinus elliottii Engelm.) stand. The site is within the Grand Bay National Wildlife Refuge and is overseen by the land managers of the refuge. The location is approximately 11.5 km from the

Gulf of Mexico, and 7.5 km northeast of the city of Pascagoula, MS (Figure 3.1). This area is approximately 9 ha in size and is the same study site used by McCoy and Keeland (2006) in their post-Hurricane Katrina evaluation. Site boundaries were set as defined in McCoy and Keeland

(2006) and through personal communication with Mr. John McCoy (Lafayette, Louisiana,

Wetland and Aquatic Research Center, U.S. Geological Survey).

The study site was partitioned into four habitats as described by McCoy and Keeland

(2006), water’s edge, levee, swamp, and slope (Figure 3.2). Water’s edge habitat was directly adjacent to the Escatawpa River or Paper Mill Canal and is approximately 0.89 hectares (ha) in size. Rising from the water’s edge is a well-defined levee that is approximately 0.71 ha of the study area. The topography then drops back down behind the levee into a swamp with standing water throughout most of the year that dominates the study site. The swamp habitat constitutes most of the site with 4.90 ha. Lastly, slope habitat was categorized as the portion of the study site that did not have year-round standing water, primarily the eastern portion of the study site

33

leading up to the longleaf and slash pine stands. The slope portion of the study area is approximately 2.55 ha.

The site falls within the Coastal Pine Meadows physiographic region of Mississippi, and is characterized by low elevation never more than 9.1 meters (m) above sea level, and little topographic relief (Lowe 1922). Three types of soil are present on the site: Maurepas muck,

Eustis loamy sand, and Axis mucky sandy clay loam which cover, 10.2% (0.9 ha), 25.3% (2.2 ha), and 64.5% (5.8 ha) of the site, respectively. Maurepas soils are very poorly drained and flood frequently. Eustis soils tend to be somewhat excessively drained with high sand components. Axis soils are classified as deep, very poorly drained, and moderately permeable

(Soil Survey Staff 2019). The National Wetlands Inventory defines the site as a freshwater forested/shrub wetland in a palustrine system due to ocean-derived salt being below 0.5 parts per thousand (ppt) (U.S. Fish and Wildlife Service 2020).

Atlantic White-Cedar Inventory

All individual Atlantic white-cedar measurements were taken following the methodology established by McCoy and Keeland (2006) and were conducted in the Fall and Winter of 2019.

The locations of all living Atlantic white-cedar stems ≥ 2.5 centimeter (cm) diameter at breast height (dbh, 1.37 m height) within the site were recorded with a GPS unit. Diameter at breast height and total heights were measured for all live Atlantic white-cedar stems. Crown class was recorded as defined by Avery and Burkhart (1983) using four categories, dominant, codominant, intermediate, and suppressed. All trees were divided into 5.08 cm (2.0 inches (in)) diameters for analysis. In order to determine the age distribution at the site, trees were randomly selected and cored from each 5.08 cm diameter size class. Two cores were taken from each tree at a 90° angle from one another at breast height. Cores were then aged by counting the number of distinct rings 34

present. If a fully intact core was not obtained, the portion of the core missing was determined by subtracting half the diameter at breast height from the length of the core that was obtained. The average width per ring was then determined for the portion of the core that was obtained and extrapolated to the missing portion of the core. Once estimated ages for each core were observed or calculated, cores were grouped into 5.08 cm diameter classes and averaged to provide an estimated average age for each 5.08 cm age class.

To assess potential competition between Atlantic white-cedar stems of Atlantic white- cedar, the average nearest neighbor tool within ArcGIS Pro 2.7.0 was utilized for all living

Atlantic white-cedar in 2019. Utilizing the aggregate points tool, Atlantic white-cedar in 2019 were grouped into patches to assess potential competition between stems of Atlantic white-cedar.

Patches were defined as three or more Atlantic white-cedar with each stem in the patch being within 6 m of another stem in the same patch. The average nearest neighbor tool was also used for only those stems that occurred within a patch. Six meters were selected for the maximum distance between Atlantic white-cedar stems for grouping patches as it is reported to be the maximum average width of a mature Atlantic white-cedar canopy (Morton Arboretum 2021).

Woody Plant Community Composition

An inventory of woody plant community composition was completed in October of 2020 to assess changes in community composition post-Hurricane Katrina and determine Atlantic white-cedar competition. Five, 10 m wide, belt transects of various lengths were re-established as defined by McCoy and Keeland (2006) and through personal communication with Mr. McCoy.

The locations and orientations of the transects were determined in 2006 to sample the variety in habitats that occur on the site. All living woody species with a dbh ≥ 2.54 cm were identified to the species level, the dbh and height were also measured for each stem. Crown class was 35

recorded as defined by Avery and Burkhart (1983). Also, 10 m2 shrub plots were established every 10 m along the transects. These shrub plots were used to quantify the level of Atlantic white-cedar seedling regeneration, competition, and community level diversity. Within each shrub plot, all woody species were identified to the species level and height was recorded. In total, 30 shrub plots were established across the site in 2020.

Transect lengths did differ slightly in length from 2006 to 2020, additionally, McCoy and

Keeland (2006) established 31 shrub plots in 2006 rather than the 30 shrub plots established in

2020. Due to these slight differences in the area sampled, all transects, and shrub plots were transformed to a per hectare basis for comparison. Each transect was expanded to a per hectare basis, then an average was calculated across the five transects for each sampling period. Results were reported for each species encountered, this method was utilized for both basal area (m2) and stems per hectare. Additionally, each shrub plot was expanded to a per hectare basis, then averaged across plots for each sampling period to estimate stems per hectare for each species.

Lastly, the Shannon-Weiner diversity index was calculated for each transect and shrub plot using the statistical software R and the R package ‘vegan’ (Barnes et al. 1998, Oksanen et al. 2019, R

Core Team 2020).

Results

Atlantic White-Cedar Inventory

In total, 871 living Atlantic white-cedar were found on the site in 2019 (Figure 3.3). The average diameter at breast height (dbh) in 2019 was 10.6 ± 0.3 cm, with a quadratic mean dbh was 14.1 cm. The greatest number of stems were in the 2.5-7.5 cm diameter class (Figure 3.4).

The average age at breast height within the smallest diameter class (2.5-7.5 cm) was 10 ± 2 years, while the average age at breast height of the largest diameter (57.51-62.5 cm) class 36

represented was 129 ± 9 years (Table 3.1). The highest density of Atlantic white-cedar stems was found within the slope habitat (235 stems/ha) (Figure 3.5). The levee (108 stems/ha), water’s edge (79 stems/ha) and swamp (26 stems/ha) had the second, third, and fourth highest densities, respectively. Few Atlantic white-cedar stems were considered dominant (27 stems), while almost half were suppressed (412 stems). Atlantic white-cedar stems on the levee had the largest average dbh (23.7 ± 1.3 cm), while those on the slope had the smallest average dbh (7.3 ± 0.2 cm) (Figure 3.5). Similarly, Atlantic white-cedar stems on the levee were on average the tallest

(13.2 ± 0.5 m), while stems on the slope were on average the shortest (8.1 ± 0.1 m). Within the shrub plots 633 ± 227 stems/ha of Atlantic white-cedar were found > 2.5 cm at dbh.

The average distance between Atlantic white-cedar’s nearest neighbor across the 9 ha site was 2.5 m in the 2019 inventory period. In total, 36 patches were identified. While the number of

Atlantic white-cedar stems within these patches was 699 or 80% of total stems on the site. The average distance between Atlantic white-cedar and its nearest neighbor for Atlantic white-cedar within patches was 1.4 m. Average dbh of Atlantic white-cedar within patches was 9.2 ± 0.3 cm.

The total land area occupied by Atlantic white-cedar patches was 2,309.1 m2. The average land area of an individual patch was 64.1 ± 44.2 m2.

Woody Plant Community Composition

Average total stems per hectare was 4,162 ± 2,132 stems/ha in 2020 (Figure 3.6). The most abundant species found within the overstory transects was swamp titi (Cyrilla racemiflora

L.) (1,265 ± 278 stems/ ha). The second and third most abundant species found on the site in

2020 were sweetgum (Liquidambar styraciflua L.) and Atlantic white-cedar (460 ± 460 stems/ ha and 357 ± 270 stems/ ha, respectively). The invasive camphortree (Cinnamomum camphora

(L.) J. Presl) and Chinese tallow tree were also found within the study site (9 ± 6 stems/ ha and 37

27 ± 12 stems/ ha, respectively). Average transect diversity was 2.07 ± 0.09. The average total overstory basal area was 49.6 ± 21.2 m2/ha in 2020 (Figure 3.6). The species with the highest basal area in 2020 were swamp tupelo and slash pine with 15.1 ± 4.4 m2/ha and 9.4 ± 3.0 m2/ha, respectively.

Within the 10 m2 shrub plots, 14,167 ± 5,650 stems/ha were found (Figure 3.6). Large gallberry (Ilex coriacea (Pursh) Chapm.) was the most prevalent species in the 2020 survey

(4,200 ± 1,416 stems/ha). The second and third most abundant species within the shrub plots in

2020 were swamp titi (4,167 ± 840 stems/ha) and red maple (Acer rubrum L.) (767 ± 491 stems/ha). Average shrub plot diversity was 0.95 ± 0.07.

Discussion

McCoy and Keeland (2006) reported 299 living Atlantic white-cedar stems on the site, with a geometric mean dbh of 25.3 cm. Compared to the 871 living Atlantic white-cedar stems and geometric mean dbh of 14.1 cm, this represents an increase of 572 stems across the study site (191%) and a decrease of 11.2 cm in average dbh (56%) between the two inventories. The increase in the total stem count of Atlantic white-cedar with a decrease in mean diameter indicate a younger, naturally regenerating stand post-Hurricane Katrina (Smith et al. 1997). However, it is not clear if this regeneration will result in a greater number of mature Atlantic white-cedar compared to pre-Hurricane Katrina. A large portion of the Atlantic white-cedar encountered in

2019 were clustered within the slope habitat type (69%). Additionally, many of these stems were clustered together in relatively small patches near tropical cyclone damage sites, the largest of which occurred on the slope in the northeast corner of the site. Most of these patches are currently in the early stages of stem exclusion, it can be expected that many of the Atlantic white-cedar within these patchers will not survive to maturity due to competition and a lack of 38

light (Oliver 1980). Tree cores taken from the Atlantic white-cedar on the site indicated substantial regeneration in the years following Hurricane Katrina. The smallest diameter class sampled showed the greatest increase in the number of stems from 2006 to 2019. The smallest diameter class also had an average breast height age of 10 years, suggesting the majority of the

Atlantic white-cedar regeneration occurred in the years directly following Hurricane Katrina.

To ascertain the net change in the number of Atlantic white-cedar stems after competitive exclusion, the area of land occupied by the patches was used since little regeneration occurred outside of these areas. Further, it is within these highly clustered patches that stem exclusion may be occurring. Mylecraine and Zimmermann (2000) report the site index for the southern range of

Atlantic white-cedar to be between 50 and 70. Therefore, a site index of 60 was used for calculating the net change in Atlantic white-cedar stems. Korstian and Brush (1931) report the number of stems per hectare of Atlantic white-cedar that are greater than 20.3 cm at 50 years with a site index of 60 to be 887 stems/ha for a fully stocked stand. Therefore, 887 stems/ha of

Atlantic white-cedar was used as the benchmark stocking for the area covered by the clustered patches of Atlantic white-cedar on the site. Using the benchmark of 887 stems per ha provided by Korstain and Brush (1931), the area covered by the patches of Atlantic white-cedar could support an estimated 205 stems at age 50. Therefore, of the 699 stems within patches observed in

2019, only 205 stems are likely to survive to age 50. Further, this would result in the estimated number of Atlantic white-cedar stems surviving 50 years in the future to be 377 stems on the site.

Compared to the 2006 survey, this would represent a net increase of 78 Atlantic white-cedar stems on the site.

Nearly half (47%) of living Atlantic white-cedar stems found within the study site were considered suppressed, and 38% were considered intermediate (Figure 3.5). It is not currently

39

clear how long the stems can remain overtopped before they begin to lose their apical dominance. It is also unclear if these stems will be able to capture available space and light if there were to be another tropical cyclone disturbance like that created by Hurricane Katrina. If these stems are no longer able to capture space and light due to a loss of apical dominance, then germination of Atlantic white-cedar from the seed bank may be critical for regenerating the species post-tropical cyclone disturbance. Further research to assess the role the seed bank and currently suppressed Atlantic white-cedar will be critical to advance understanding of the regeneration of Atlantic white-cedar after tropical cyclone disturbance events.

Due to a lack of seedlings and saplings in 2006, McCoy and Keeland (2006) reported concerns for the sustainability of the species within the study site. In 2006, McCoy and Keeland only reported 15 Atlantic white-cedar seedlings/ha, compared to the 633 ± 227 seedlings/ha found after Katrina, this represents a 4,120% increase in Atlantic white-cedar seedlings/ha on the study site. McCoy and Keeland (2006) suggested that the canopy gap along the slope portion of the study site may have been opened enough by Hurricane Katrina to support the future natural regeneration of Atlantic white-cedar. By 2019, Atlantic white-cedar had regenerated prolifically on this site, particularly on the northeast portion of the slope. However, as previously discussed, regeneration was primarily within relatively small patches that might ultimately limit the number of Atlantic white-cedar reaching maturity due to intraspecies competition. Further regeneration of Atlantic white-cedar seedlings and saplings will need to occur outside these patches to prevent further intraspecies competition and fully utilize the site. The increase of Atlantic white-cedar seedlings found within shrub plots is likely due to light reaching the forest floor. Immediately after Hurricane Katrina, canopy gaps allowed sunlight to penetrate the forest floor, in the years since these gaps have been closed by competing vegetation, predominantly large gallberry,

40

swamp titi, red maple, and thick patches of larger Atlantic white-cedar. The increase in the number of seedlings indicates a potentially favorable outlook for further Atlantic white-cedar regeneration on the site, if the seedlings can outcompete other vegetation.

Competing vegetation increased substantially between 2006 and the 2020 inventories.

McCoy and Keeland (2006) reported 2,375 overstory stems/ha in 2006, compared to the 2020 inventory, there was a 75% increase in overstory stems between inventories. Further, McCoy and

Keeland (2006) reported less than 163 stems/ha within the shrub plots, compared to the 2020 inventory, this represents an increase of 8,591% between inventories. While Atlantic white-cedar seedlings increased, theses seedlings could be soon outcompeted and die out by the increase in midstory and overstory vegetation. Further, the increase in competing overstory stems may hinder the development of larger Atlantic white-cedar through intraspecies competitive exclusion. Silvicultural practices, such as herbicide treatments, may be needed to hamper the development of competing species for Atlantic white-cedar to regenerate and perpetuate on the site (Laing et al. 2011).

Laing et al. (2011) found that Atlantic white-cedar required active management to effectively regenerate post- tropical cyclone disturbance within the Great Dismal Swamp of

North Carolina and Virginia, USA. In areas where intensive management was implemented,

Atlantic white-cedar seedling densities were 100 times higher compared to areas of passive management. Examples of intense management described by Laing et al. (2011) include debris removal and herbicide treatments to reduce competition and increase available growing space.

No active management is known to have occurred for the restoration of Atlantic white-cedar within the Grand Bay National Wildlife Refuge study site. It is currently unclear if the active management and results reported by Laing et al. (2011) apply to the more species diverse stands

41

containing Atlantic white-cedar within the Gulf Coast region. However, due to the increases in competing mid- and overstory vegetation, it is possible active and intentional management, as described by Laing et al. (2011), could increase recruitment of Atlantic white-cedar seedlings into larger size classes.

Of notable concern were several invasive species found on the site. The most prevalent invasive found, the Chinese tallow tree, is a common invader of Southeastern wetlands (Jubinsky and Anderson 1996). Chinese tallow tree has been shown to outperform native Southeastern species in greenhouse studies that simulated post- tropical cyclone conditions, such as increased light and increased water salinity (Paudel and Battaglia 2020). Further, Chinese tallow tree was reported to increase substantially following Hurricane Katrina in a bottomland hardwood forest system in Louisiana, USA (Henkal et al. 2016). McCoy and Keeland did not report any Chinese tallow within the study site in 2006. The increase in Chinese tallow stems from 2006 to 2020 indicates the species is spreading within the study site, potentially because of disturbance created by Hurricane Katrina. Additionally, camphortree spread within the study site from 2006 to 2020, albeit less than Chinese tallow tree. McCoy and Keeland (2006) reported 3 stems/ha of camphor tree in 2006. The role of disturbance, and particularly tropical cyclone in the regeneration of camphortree is unclear. However, such a small increase in the number of stems from 2006 to

2020 could indicate camphortree did not benefit as well from the tropical cyclone disturbance, as did Chinese tallow tree. Although not recorded in the woody plant community composition inventory, Japanese climbing fern (Lygodium japonicum (Thunb.) Sw.) was also prevalent within the study site. This species is known to form thick mats on the forest floor that can prevent the regeneration of native plant species. Disturbance, particularly windstorms, has been shown to promote the invasion of Japanese climbing fern (Carmichael 2012). The establishment and

42

spread of these invasive species on the study site could prove detrimental to future Atlantic white-cedar and other native plants within the community studied.

After 14 years, the effects of Hurricane Katrina on the Atlantic white-cedar at the Grand

Bay National Wildlife Refuge have yet to be fully realized. Due to this, and the variability in plant communities containing Atlantic white-cedar on the Gulf Coast, specific land management recommendations for the conservation and regeneration of Atlantic white-cedar are difficult to establish. Yet, some generalities can be drawn from the regeneration of Atlantic white-cedar at

Grand Bay National Wildlife Refuge. These general recommendations do align with the findings of Laing et al. (2011), that active management may be required to support Atlantic white-cedar regeneration post- tropical cyclone disturbance. Mechanical or chemical control of understory hardwood species, such as large gallberry and swamp titi, may be needed to promote Atlantic white-cedar recruitment into larger size classes. Such efforts could allow more light to reach

Atlantic white-cedar seedlings currently under relatively thick midstory vegetation. Additionally, invasive species management may be necessary to retain the native flora in the current community.

Conclusions

The number of Atlantic white-cedar stems increased substantially since McCoy and

Keeland first surveyed the populations. However, active management, such as mid-story hardwood removal and invasive species control may be needed to aid regeneration and facilitate the long-term success of Atlantic white-cedar within the study site. This study also provides the evidence that tropical cyclones, not fire, may be contributing to the regeneration of Atlantic white-cedar on the Gulf Coast. Lastly, this study provides land managers and conservationist

43

insights into how Atlantic white-cedar and their associated community may regenerate long-term after a tropical cyclone disturbance.

Acknowledgements

We thank Mr. John McCoy and the Wetland and Aquatic Research Center at the U.S.

Geological Survey for their support. This work would not have been possible without a team of dedicated graduate and undergraduate colleagues who assisted with data collection, Will

Kruckeberg, Spencer Madden, Darcey Collins, Connor McClendon, Hunter Kelly, Austin Reese,

Baylor Doughty, Harris Groberg, and Garrett Hendrix. We would also like to thank Scott

Hereford and the staff of the Grand Bay National Wildlife Refuge for their continued support of this research and access to the study site. Lastly, we would like to thank the staff of the Grand

Bay National Estuarine Research Reserve, particularly Sandra Huynh, for their logistical support of this project, without which, this project would not have been possible. This work was supported by the National Institute of Food and Agriculture, U.S. Department of Agriculture,

Mclntire Stennis capacity grant #1019117. Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the author(s) and do not necessarily reflect the view of the U.S. Department of Agriculture. S. B. Correa was supported by

Mississippi State University (# MISZ-081700). This publication is a contribution of the Forest and Wildlife Research Center, Mississippi State University.

44

Tables

Table 3.1 Age at breast height of Atlantic white-cedar stems within the study site

Diameter Class (cm) N Mean Age (years)

2.5-7.5 9 10 (1)

7.51-12.5 16 26 (4)

12.51-17.5 6 46 (5)

17.51-22.5 7 37 (2)

22.51-27.5 12 49 (3)

27.51-32.5 12 64 (4)

32.51-37.5 9 74 (4)

37.51-42.5 10 77 (7)

42.51-47.5 4 102 (12)

47.51-52.5 0 NA

52.51-57.5 3 87 (3)

57.51-62.5 2 129 (7) Average age at breast height by size class for Atlantic white-cedar (Chamaecyparis thyoides (L.) B.S.P.) within the 9 ha study site at the Grand Bay National Wildlife Refuge in Jackson County, Mississippi, USA in 2019. Standard errors are presented parenthetically.

45

Figures

Figure 3.1 Location of Atlantic white-cedar study site in Southern Mississippi, United States.

Study site located within the Grand Bay National Wildlife Refuge in Jackson County, Mississippi, USA in relation to the Southeastern United States with a black star representing the location of the study site.

46

Figure 3.2 Map of Atlantic white-cedar study site in Southern Mississippi, United States

The four habitat types defined by McCoy and Keeland (2006) within the 9 ha study site at the Grand Bay National Wildlife Refuge in Jackson County, Mississippi, USA.

47

Figure 3.3 Locations of all Atlantic white-cedar stems found within the study site in Southern Mississippi, United States

All Atlantic white-cedar (Chamaecyparis thyoides (L.) B.S.P.) within the 9 ha study site at the Grand Bay National Wildlife Refuge in Jackson County, Mississippi, USA in 2019.

48

Figure 3.4 Diameter distribution of all Atlantic white-cedar stems within the study site in Southern Mississippi, United States

Diameter distribution for Atlantic white-cedar (Chamaecyparis thyoides (L.) B.S.P.) within the 9 ha study site at the Grand Bay National Wildlife Refuge in Jackson County, Mississippi, USA in 2019.

49

Figure 3.5 Descriptive analyses of Atlantic white-cedar within the study site by crown class and generalized habitat in Southern Mississippi, United States

(a) The number of Atlantic white-cedar (Chamaecyparis thyoides (L.) B.S.P.) ≥ 2.5 cm at dbh within four crown classes within the 9 ha study site at the Grand Bay National Wildlife Refuge in Jackson County, Mississippi, USA. (b) The number of Atlantic white-cedar stems within the four different generalized habitat types found within the study site. (c) A boxplot of the distribution of the dbh of Atlantic white-cedar within the four different generalized habitat types found within the study site. (d) A boxplot of the distribution of the total height of Atlantic white- cedar within the four different generalized habitat types within the study site. 50

Figure 3.6 Competing overstory and midstory vegetation within the Atlantic white-cedar study site in Southern, Mississippi, United States

(a) Average stems/ ha of overstory tree species (≥ 2.5 cm at dbh) within the 9 ha study site at the Grand Bay National Wildlife Refuge in Jackson County, Mississippi, USA in 2020 (b) Average basal area in m2/ha of overstory tree species within the Grand Bay National Wildlife Refuge Study Site and (c) Average stems/ha of woody species with a dbh < 2.5 cm within the Grand Bay National Wildlife Refuge Study Site. 51

Literature Cited

Avery, T.E., and H.E. Burkhart. 1983. Forest Measurements. McGraw-Hill. Boston, MA.

Barnes, B.V., D.R. Zak, S.R. Denton, and S.H. Spurr. 1998. Forest Ecology. 4th ed. Wiley, Hoboken, New Jersey.

Camargo, S.J. and A.A. Wing. 2021. Increased tropical cyclone risk to coasts. Science 371(6528):458-459.

Carmichael, B.J. 2012. Effects of fire and treefalls on Japanese climbing fern and native species ground cover in restored longleaf pine savanna. Doctoral Dissertation. Louisiana State University. Baton Rouge, Louisiana.

Comita, L.S., M. Uriarte, J. Thompson, I. Jonckheere, C.D. Canham, and J.K. Zimmerman. 2009. Abiotic and biotic drivers of seedling survival in a hurricane‐impacted tropical forest. Journal of Ecology 97(6):1346-1359.

Dale, V.H., L.A. Joyce, S. McNulty, R.P. Neilson, M.P. Ayers, M.D. Flannigan, P.J. Hanson, L.C. Irland, A.E. Lugo, C.J. Peterson, D. Simberloff, F.J. Swanson, B.J. Stocks, and B.M Wotton. 2001. Climate change and forest disturbances: climate change can affect forests by altering the frequency, intensity, duration, and timing of fire, drought, introduced species, and pathogen outbreaks, hurricanes, windstorms, ice storms, or landslides. BioSience. 51(9):723-734.

Doyle, T.W. 2009. Hurricane frequency and landfall distribution for coastal wetlands of the Gulf Coast, USA. Wetlands. 29(1):35-43.

Ehrenfeld, J.G., and J.P. Schneider. 1991. Chamaecyparis thyoides wetlands and suburbanization: Effects on hydrology, water quality and plant community composition. Journal of Applied Ecology. 28:467-490.

Eleuterius, L.N., and S. B. Jones. 1972. A phytosociological Study of white-cedar in Mississippi. Castanea. 37:67-74.

Elsner, J.B., and T.H. Jagger. 2010. On the increasing intensity of the strongest Atlantic hurricanes. Pp. 175-190. in J. Elsner, R. Hodges, J. Malmstadt, and K. Scheitlin, ed., Hurricanes and climate change, Springer, Dordrecht, Netherlands.

Everham, E.M. and N.V.L. Brokaw. 1996. Forest damage and recovery from catastrophic wind. The Botanical Review 62:113-185.

Hardin, J.W., F.M. White, and D.J. Leopold. 2000. Harlow & Harrar’s Textbook of Dendrology. McGraw-Hill. Boston, MA.

52

Henkal, T.K., J.Q. Chambers, and D.A. Baker. 2016. Delayed tree mortality and Chinese tallow (Triadica sebifera) population explosion in a Louisiana bottomland hardwood forest following Hurricane Katrina. Forest Ecology and Management 378:222-232.

Ibañez, T., G. Keppel, C. Menkes, T.W. Gillespie, M. Lengaigne, M. Mangeas, G. Rivas-Torres, and P. Birnbaum. 2019. Globally consistent impact of tropical cyclones on the structure of tropical and subtropical forests. Journal of Ecology 107(1): 279-292.

Jubinsky, G. and L.C. Anderson. 1996. The invasive potential of Chinese tallow-tree (Sapium sebiferum Roxb.) in the southeast. Castanea. 61(3) 226-231.

Keeland, B.D., and J.W. McCoy. 2007. Plant community composition of a tidally influenced, remnant Atlantic white cedar stand in Mississippi. Pp. 89-112. in W.H. Conner, T.W. Doyle, and K.W. Krauss, ed., Ecology of Tidal freshwater forested wetlands of the Southeastern United States, Springer, Dordrecht, Netherlands.

Korstian, C.F. 1924. Natural regeneration of southern white cedar. Ecology. 5:188-191.

Korstian, C.F., and W.D. Brush. 1931. Southern white cedar. U.S. Department of Agriculture. Technical Bulletin 251.

Laderman A.D. 1989. The ecology of Atlantic white cedar wetlands: a community profile. U.S. Fish and Wildlife Service Biological Report. 85(#7.21).

Laing, J.M., T.H. Shear, and F.A. Blazich. 2011. How management strategies have affected Atlantic white-cedar forest recovery after massive wind damage in the Great Dismal Swamp. Forest Ecology and Management. 262:1337-1344.

Li, H.L. 1962. A new species of Chamaecyparis. Morris Arboreteum Bullatin. 13:43-46.

Lin, T.C., J.A. Hogan, and C.T. Chang. 2020. Tropical Cyclone Ecology: A Scale-Link Perspective. Trends in Ecology and Evolution 35(7):594-604.

Little, E.L. 1966. Varietal transfers in Cupressus and Chamaecyparis. Madrono. 18:161-167.

Little, E.L. 1979. Checklist of United States Trees. U.S. Department of Agriculture Forest Service. Agricultural Handbook 541.

Lowe, E.N. 1922. Plants of Mississippi. Botanical Gazette. 73:422-422.

Lugo, A. E. 2000. Effects and outcomes of Caribbean hurricanes in a climate change scenario. Science of the Total Environment. 262(3):243-251.

Lugo, A.E., and F.N. Scatena. 1996. Background and Catastrophic Tree Mortality in Tropical Moist, Wet, and Rain Forests. Biotropica 28(4):585-599.

53

McCoy, J.W., and B.D. Keeland. 2006. Species composition and hurricane damage in an Atlantic white cedar stand near the Mississippi/Alabama border. Pp. 8-20 in Proceedings of the Ecology and Management of Atlantic White-Cedar Symposium, 6-8 June, 2006, Atlantic City, New Jersey.

Morton Arboretum. 2021. Atlantic white-cedar. Available online . Accessed 1 January 2021.

Mylecraine, K.A., and G.L. Zimmermann. 2000. Atlantic white-cedar ecology and best management practices manual. Division of Parks and Forestry, New Jersey Forest Service, Trenton, New Jersey.

NatureServe Explorer. 2020. An Online Encyclopedia of Life: Atlantic White-cedar. Available at . Accessed 26 April, 2020.

Nowacki, G.J., and M.D. Abrams. 2008. The demise of fire and “mesophication” of forests in the eastern United States. Bioscience. 58(2):123-138.

Oksanen, J., F.G. Blanchet. M. Friendly, R. Kindt, P. Legendre, D. McGlinn, P.R. Minchin. R.B. O’Hara, G.L. Simpson, P. Solymos, M. Henry, H. Stevens, E. Szoecs, and H. Wagner. 2019. Vegan: Community ecology package. R Package Version 2.5-6.

Oliver, C.D. 1980. Forest development in North America following major disturbances. Forest Ecology and Management. 3:153-168.

Patricola, C.M. and M.F. Wehner. 2018. Anthropogenic influences on major tropical cyclone events. Nature 563:339-346.

Paudel, S. and L.L. Battaglia. 2020. Linking responses of native and invasive plants to hurricane distrubances: implications for coastal plant community structure. Plant Ecology. 222:133- 148.

Smith, D.M., B.C. Larson, M.J. Kelty, and M.S. Ashton. 1997. The practice of silviculture: applied forest ecology. Wiley. Hoboken, New Jersey. P. 189-218.

R Core Team. 2020. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. < https://www.r-project.org/> Accessed 22 August 2021.

Soil Survey Staff. 2019. Web Soil Survey. Natural Resources Conservation Service, U.S. Department of Agriculture, Washington, D.C. Accessed 22 August 2021

54

U. S. Fish and Wildlife Service. 2020. National Wetlands Inventory website. Fish and Wildlife Service, U.S. Department of the Interior, Washington, D.C. Accessed 22 August 2021.

Wang, S., and R. Toumi. 2021. Recent migration of tropical cyclones toward coasts. Science 371(6528):514-517.

Ward, D.B. 1989. Commercial utilization of Atlantic white cedar (Chamaecyparis thyoides, Cupressaceae). Economic Botany. 43:386-415.

Ward, D.B. and A.F. Clewell. 1989. Atlantic white cedar (Chamaecyparis thyoides) in the southern states. Florida Scientist. 52(1):8-47.

Webster, P.J., G.J. Holland, J.A. Curry, and H.R. Chang. 2005. Changes in tropical cyclone number, duration, and intensity in a warming environment. Science. 309(5742):1844- 1846.

Xi, W. 2015. Synergistic effects of tropical cyclones on forest ecosystems: a global synthesis. Journal of Forestry Research 26:1-21.

55

CHAPTER IV

MICROSITE HABITAT, SPECIES ASSOCIATIONS, AND HABITAT SUITABILITY

MODEL OF A GLOBALLY IMPERILED SHRUB

Abstract

Miller’s witch-alder (Fothergilla milleri, Hamamelidaceae) is a newly described globally imperiled shrub that is known from disjunct populations in coastal Alabama, the panhandle of

Florida, and one location in Georgia. Little is currently known about the natural history or ecology of the species. Across the three populations sampled, 3,060 stems were found, 45% were in one patch ≤ 0.5 hectares (ha) in area and 79% were in one population within an area of 4.5 ha.

Low stem counts in few locations makes the species particularly vulnerable to extinction due to stochastic events. Also, all seed pods found within the surveys were infected by an unknown aphid, which is likely preventing sexual reproduction within the populations. Microsite habitat data indicated F. milleri grows in a unique transitional habitat between upland conifer forest and wetlands. The sites where extant populations of F. milleri inhabit tend to have acidic (4.860 ±

0.081) and well-drained soils with a high sand content. Further, many populations were under thick midstory vegetation. Thus, fire or other forms of removal may be needed to release those patches before they may become shaded out. Lastly, only 0.5% of the study area within the habitat suitability model was identified within the highest level of suitability. Fothergilla milleri is facing multiple threats that could lead to its extirpation from the wild, and direct and intense conservation action may be necessary to ensure F. milleri remains on the landscape.

56

Introduction

Miller’s witch-alder (Fothergilla milleri W. D. Phillips & J. E. Haynes, Hamamelidaceae) is a small shrub endemic to the Gulf Coast Plain of Alabama, the panhandle of Florida, and one county in Georgia (Haynes et al. 2020, Figure 4.1). Fothergilla milleri is most similar morphologically to dwarf witch-alder (Fothergilla gardenii L.) and was not identified as a unique species until 2020 (Haynes et al. 2020). The largest morphological differences between these two species are the orientation of the leaves; F. milleri leaves tend to be held erect, while

F. gardenii leaves tend to be spreading (Haynes et al. 2020). Additionally, F. milleri varies genetically from F. gardenii. Fothergilla milleri presents diploid chromosomes while F. gardenii presents tetraploid chromosomes (Haynes et al. 2020). According to genetic research conducted by Haynes et al. (2020), the only other diploid species in the genus is F. parvifolia (Kearney). F. parvifolia and F. milleri differ by geographical distribution with Fothergilla parvifolia occurring primarily in eastern Georgia and South Carolina and tend to have leaves that droop along the stems (Haynes et al. 2020). Beyond morphological descriptions and brief habitat generalizations, little is known about the newly described F. milleri. This large knowledge gap in the ecology, habitat needs, and current conservation statutes severely limits conservationists' ability to protect and manage for the species.

Fothergilla milleri has been ranked as a G2 globally imperiled species by NatureServe due to several possible threats (Natureserve Explorer 2021). The lack of naturally occurring fire regimes have been considered one factor leading to the decline F. gardenii populations across their natural range (Barger et al. 2013, Haynes et al. 2020). Due to similarities in habitat and distribution, fire suppression is likely a factor impacting F. milleri abundance as well. With so few known populations, another threat currently facing F. milleri is environmental stochasticity.

57

While moderate disturbance may support regeneration of F. milleri, one extreme environmental event, such as a severe fire or tropical cyclone, could potentially reduce or eliminate extant populations of F. milleri (Menges 1992). Fothergilla milleri likely faces several threats, however, little is known of these threats.

Modeling the habitat suitability of rare and imperiled plant species has become an integral part of plant conservation (Kumar and Stohlgren 2009, Yang et al. 2013, Remya et al.

2015, Sharma et al. 2018, Ramirez-Reyes et al. 2021). Habitat suitability models allow plant conservationists to more effectively allocate resources to areas that are more likely to support the species of concern. Additionally, habitat suitability models can concentrate survey efforts on habitats that are most likely to support the species (Costa et al. 2010). Several methodologies are available to produce habitat suitability models, however, one of the more popular methods is maximum entropy, or Maxent (Philips et al. 2006). Maxent seems to perform particularly well for species with small sample sizes, like that of F. milleri, due to the robustness of the model

(Hernandez et al. 2006, Wisz et al. 2008). Habitat suitability models can therefore aid conservation activities to help ensure the perpetuation of rare and imperiled plant species on the landscape (Hemati et al. 2020).

The main objectives of this study were to: (1) assess the conservation status of the three largest known populations of F. milleri, (2) quantify the microhabitats and species associations where F. milleri occurs, and (3) provide a preliminary habitat suitability map to assist conservationists with future decision making. The last two objectives were designed to facilitate the identification of suitable areas for F. milleri establishment. Results of this model open the possibility of out planting, or introducing, F. milleri into areas with high habitat suitability.

Fothergilla milleri is not currently widely available in the nursery trade, however, F. gardenii is

58

(Ranney et al. 2007). Understanding the habitat needs of F. milleri in the wild, particularly soil conditions and optimal light conditions, may facilitate horticulturalists in bringing F. milleri into cultivation.

Methods

Sampling Areas

Through personal communication Ron Miller1, we identified the three largest known populations of F. milleri (Figure 4.2). One smaller populations of F. milleri is known, however, it was not used due to a lack of access and small population size. The first population inventoried is within the Perdido Wildlife Management Area, approximately 5 kilometers (km) east of

Gateswood in Baldwin County, Alabama, USA. The second population is approximately two kilometers west of the city of Geneva in Geneva County, Alabama, USA. The third population is located approximately ten kilometers northeast of the city of Crestview in Okaloosa County,

Florida, USA. Throughout this publication, the populations will be referred to by the county in which they are located (Baldwin County, Geneva County, and Okaloosa County).

Stem Counts

A 100% inventory of F. milleri was conducted in all populations, except one patch in the

Okaloosa County population. One patch within the Okaloosa population was subjected to a prescribed burn before a 100% survey could be conducted. A preliminary survey using a modified point-centered quarter, or Wisconsin1 plot, method had been conducted before the prescribed burn, therefore, for this one patch, an estimated number of F. milleri stems was used

(Cottam and Curtis 1956). For this one patch within the Okaloosa population, a 20 m transect

1 Ron Miller of Pensacola, Florida is the local species expert and namesake of the species. 59

was established through the population. Three points, one at either end and one in the middle, were then established. At each point, four 0.25 meters squared (m2) quadrats were placed at a random distance between 1 m and 5 m away from the point at 45°, 135°, 225°, and 315° offsets from the transect. All F. milleri stems were counted within the quadrat then summed for each of the four points. The results of the transect were then extrapolated to the total size of the patch.

For all other patches within the Baldwin, Geneva, and Okaloosa populations a 100% systematic survey was carried out. Due to the rhizome growth form of F. milleri differentiating between genetic individuals would be difficult to nearly impossible without extensive genetic analysis of each stem. Therefore, individual stems were counted regardless if they were genetic individuals or not. A stem was defined at the soil level where F. milleri was growing from a single point.

Lastly, the number of seed pods that appeared healthy and uninfected was noted throughout the survey period.

Woody Vegetation Sampling

Within the Baldwin and Okaloosa County populations, the distribution of F. milleri was patchy and noncontiguous. For each patch of F. milleri located within each population, a woody vegetation sampling plot was established in the center of the patch. In total, nine woody vegetation sampling plots were established within the Baldwin County population and five were established within the Okaloosa County population. The population in Geneva County was a single ~30 m long patch along Goat Hill Road. Two woody vegetation sampling plots were established within the Geneve County populations, one at either end of the population along the road.

Overstory species association was determined in fixed-area plots 0.05 hectare (ha) in size

(Figure 4.3). All woody stems greater than 10 cm at breast height (1.4 meters (m)) were 60

identified to the species level and diameter at breast height (dbh) was measured. Midstory species association plots were 0.025 ha and were nested within each overstory plot. All woody stems with diameters between 2.5 cm and 10 cm at breast height were identified to the species level. Two diameter measurements were taken at breast height using handheld calipers for each stem within the midstory plots, the two measurements were then averaged for each stem dbh.

To determine maximum overstory tree height, the three tallest canopy stems per plot were measured with a Suunto Clinometer or TruPulse 200 laser clinometer. The three tree heights were then averaged to estimate the average overstory tree height for each plot. To estimate canopy cover, a Model-c spherical concave densiometer produced by Forest Densiometers was used. Canopy cover measurements were taken at the four cardinal directions along the perimeter of the midstory vegetation plots. At each of the four cardinal directions along the midstory vegetation plot perimeter, one measurement was recorded facing each of the four cardinal directions. This resulted in a total of 16 canopy cover measurements per plot. These measurements were then averaged to produce one average canopy cover estimate per plot.

A one-way ANOVA was used to test if results were consistent among the three populations among the three populations in overstory stems/ha, midstory stems/ha, overstory basal area, midstory basal area, overstory height, and canopy cover. If statistical differences were detected, it could indicate differing ecological niches among populations. Both midstory stems/ha and midstory basal area were not normally distributed. Therefore, a square root transformation was conducted to meet the normal distribution assumption. If statistically significant differences were detected among populations a Tukey’s pair-wise comparison was then used to determine which populations were different from one another. Additionally, simple linear regression models were developed to test for a correlation between the number of F.

61

milleri stems and the previously listed factors. All tests were conducted with the statistical software R and an alpha of 0.05 was used to test for significance (R Core Team 2020).

Soil Sampling

Soil samples were taken at each of the vegetation plots at the plot center. A 30 centimeters (cm) long soil push probe with a two-centimeter-wide opening was used to collect each soil sample. The sample was divided into two separate samples, the top 10 cm of soil and the remaining bottom 20 cm. Samples were then placed in resealable plastic bags and placed on ice and later transferred to refrigeration. Soil texture for each sample was determined by feel

(Thien 1979). Soil pH was measured using a HANNA HI 5521 Research Grade pH/EC meter with 0.001 accuracy and a 2:1 slurry ratio. The soil C:N ratio was measured using a ECS 4010

CHNOS Elemental Analyzer. The U.S. Natural Resource Conservation Science Web Soil Survey and U.S. Fish and Wildlife Service National Wetlands Inventory were both used to further characterize soil conditions, such as texture and distance to wetlands (Soil Survey Staff 2021,

U.S. Fish and Wildlife Service 2020).

A one-way ANOVA was used to test for differences among the three populations in soil pH and C:N ratio. If statistically significant differences were detected among populations a

Tukey’s pair-wise comparison was then used to determine which populations were different from one another. Soil C:N ratio was found not to meet the normal distribution assumption; therefore, a common logarithmic transformation was used to meet the assumption. Additionally, simple linear regression models were developed to test for a correlation between the number of F. milleri stems and soil pH, soil layer depths, and carbon to nitrogen ratios. All tests were conducted with the statistical software R and an alpha of 0.05 was used to test for significance (R

Core Team 2020). 62

Habitat Suitability Model (HSM)

A preliminary habitat suitability model for F. milleri is also presented. Maxent version

3.4.0 was used to develop the model (Phillips et al. 2006). HSMs utilize occurrence records of a species and environmental raster layers to estimate suitable habitat for that species across a defined area. Unlike other HSM methodologies, Maxent utilizes presence-only occurrence data rather than presence-absence data (Phillips et al. 2006, Phillips and Dudík 2008). Opportunistic botanical surveys can therefore be used in the modeling process, as opposed to formal systematic surveys. The study area for the presented model was an ellipse that encompasses all the occurrences used in the modeling process. An ellipse was used around the known occurrences to prevent the model from extrapolating beyond those points. The defined study area was set to encompass approximately 114,115 km2.

To optimize the presented model, the methodologies of Shcheglovitova and Anderson

(2013) were followed. Shcheglovitova and Anderson (2013) combined the methodologies of

Anderson and Gonzalez (2011) and Syfert et al. (2013) to optimize both the regularization multiplier and feature class in a combination of one another for use within the final model. These methodologies are designed specifically for modeling applications when there are less than 25 available occurrence points. Various regularization multipliers, as defined by Shcheglovitova and

Anderson (2013), were used in combination with various feature classes, or combinations of feature classes. Regularization, or beta, multipliers of 0.5, 0.75, 1.0, 1.25, 1.75, and 2 were used in model optimization. The feature classes and feature class combinations used were linear (L), hinge (H), linear with quadratic (LQ), and linear with quadratic and hinge (LQH). With the six regularization multipliers and four feature classes, a total of 28 models were developed. A 63

jackknife, or leave one out, approach was used for each model. The lowest presence threshold rule, or minimum training presence threshold in Maxent, was used to calculate the binary omission rate (Pearson et al. 2007, Shcheglovitova and Anderson 2013) and the optimal model was selected based on the lowest omission rate (OR). Through this process the combination of a regularization multiplier of one and the linear feature class was found to be optimal and was therefore used in the final model. Background data, or pseudo-absences, were limited to hydrological unit code (HUC) 8 level watersheds that contain one of the eight F. milleri occurrences used within the model (U.S. Geological Survey 2010). A uniform prior was therefore assumed within the watersheds (Merow et al 2013). After the modeling process within the watersheds, the model was then projected out to the ellipse study area. In total 5,000 background points were used, with all other settings within the Maxent software set on default.

Lastly, a jackknife test was used to determine the importance of the various environmental predictor levels within the final model.

Initially, 71 occurrences were considered with eight occurrences being selected for use in the final model. GPS coordinates for the occurrences were provided by Ron Miller. To avoid autocorrelation, occurrence data was rarified to 2 km2, which was the original size of the largest raster cells. The rarefication tool within the SDMtoolbox was used to accomplish the rarefication

(Brown et al. 2017). Sixty-eight occurrences were excluded due to the spatial rarefication process. After the rarefication process, eight occurrences were used in the final model.

Twenty of the 22 originally considered environmental predictor variables were used in the final model. ArcGIS Pro 2.4.3 was used to prepare all environmental layers. Similar to the methodologies of Shcheglovitova and Anderson (2013), we initially used 19 bioclimatic variables within the WorldClim dataset (Fick and Hijmans 2017). The 19 bioclimatic variables

64

considered were; annual mean temperature (bio1), mean diurnal range (bio2), isothermality

(bio3), temperature seasonality (bio4), maximum temperature of the warmest month (bio5), minimum temperature of the coldest month (bio6), temperature annual range (bio7), mean temperature of the wettest quarter (bio8), mean temperature of the driest quarter (bio9), mean temperature in the warmest quarter (bio10), mean temperature of the coldest quarter (bio11), annual precipitation (bio12), precipitation in the wettest month (bio13), precipitation in the driest month (bio14), precipitation seasonality (bio 15), precipitation of the wettest quarter (bio16), precipitation of the driest quarter (bio17), precipitation of the warmest quarter (bio18), precipitation of the coldest quarter (bio19). However, the mean temperature of the wettest quarter (bio8) and the mean temperature of the driest quarter (bio9) were dropped from the final model. Their inclusion in the final model produced bands within the map that were inappropriate at the scale of this model. These bioclimatic datasets are biologically relevant climatic data indices that are calculated using standard climate data from 1970-2000 (Fick and Hijmans 2017).

Worldclim’s bioclimatic data has been used successfully to model other rare plant species (Yang et al. 2013, Abdelaal 2019, Remirez-Reyes et al. 2021). The 30 second, or ca. 1 km2, resolution was used for all WorldClim environmental predictor layers. In addition to the 17 bioclimatic datasets used in the final model, three other environmental predictor variables were developed and used within the final model. Due to anecdotal evidence of F. milleri growing near wetlands reported by conservationist, the Euclidean distance from a body of water was one of these additional datasets developed. To create this layer the USA Detailed Streams layer was rasterized and used within the Euclidean distance function in ArcGIS Pro (Gary et al. 2009). The two remaining environmental predictor layers were soil pH and percent soil content, both were downloaded from the SSURGO database at 30 m resolution (Soil Survey Staff 2019).

65

After the final model was developed, the map produced by Maxent was further analyzed.

Each pixel within the habitat suitability map was binned into discrete classes for each tenth of habitat suitability. The 0.9-1 habitat suitability bin is considered to represent habitat that would be of the highest suitability for the species.

Results

Stem Counts

A total of 3,060 F. milleri stems were counted across the three populations (Figure 4.4).

The Baldwin County population contained the largest number of stems with 2,425 stems in total.

The Okaloosa County population had a total of 503 stems. The one patch within the Okaloosa

County population of F. milleri that could not be fully counted due to the prescribed burn had an estimated 200 ± 116 stems. In total, 102 stems were found at the Geneva County population. One patch of F. milleri within the Baldwin population accounted for 1,367 stems, or 45% of all stems counted across the populations. Lastly, no healthy, uninfected seed pods were found throughout the sampling period.

Woody Vegetation Sampling

The average number of total overstory stems per hectare across all woody vegetation sampling plots was 484 ± 218 stems/ha. The most common species within the overstory plots containing F. milleri were loblolly pine (Pinus taeda L.), representing 33.3% of overstory stems

(161 ± 57 stems/ha, Table 4.1). The second and third most common species were longleaf pine

(Pinus palustris Mill.) and Atlantic white-cedar (Chamaecyparis thyoides (L.) Britton, Sterns &

Poggenb.) (80 ± 36 stems/ha and 79 ± 41 stems/ha, respectively). The total average basal area within the overstory vegetation plots was 18.4 ± 8.1 m2/ha. The species with the highest average

66

basal area per hectare was loblolly pine (5.87 ± 2.0 m2/ha). The second and third species with the highest basal area per hectare were longleaf pine and slash pine (Pinus elliottii Engelm.) (4.47 ±

2.1 m2/ha and 2.95 ± 1.2 m2/ha, respectively). No statistically significant differences were detected between populations with respect to both overstory stems/ha and basal area (p = 0.318 and p = 0.352, respectively).

The average number of total midstory stems per hectare across all woody vegetation sampling plots was 584 ± 307 stems/ha (Table 4.1). The most common species within midstory vegetation plots was buckwheat tree (Cliftonia monophylla (Lam.) Britton ex Sarg) (213 ± 88 stems/ha). The second and third most common species within the midstory were large gallberry

(Ilex coriacea (Pursh) Chapm) and sweetbay magnolia (Magnolia virginiana L.) (101 ± 65 stems/ha and 64 ± 29 stems/ha, respectively). The average total basal area within the midstory was 1.2 ± 0.7 m2/ha. The species with the highest basal area on average was buckwheat tree (0.5

± 0.3 m2/ha). The species with the second and third most basal area on average in the midstory were Atlantic white-cedar and large gallberry (0.2 ± 0.1 m2/ha and 0.1 ± 0.1 m2/ha, respectively).

Both midstory stems/ha and midstory basal area varied significantly between population (p =

0.001 and p = 0.012, respectively). With respect to stems/ha and basal area within the midstory, the Baldwin population significantly differed from the Okaloosa population (p = 0.001 and p =

0.001, respectively). No other statistically significant differences were found between populations.

The average maximum canopy tree height across all populations was 20.5 ± 0.6 m. No statistically significant differences were found between populations with respect to average maximum canopy height (p = 0.341). The average canopy cover across all plots was 77.3 ±

3.1%. The Geneva population had the lowest average canopy cover (60.9 ± 0.7%). The Okaloosa

67

population had the second-lowest average canopy cover (70.0 ± 3.9%). Lastly, the Baldwin population had the highest canopy cover (85.1 ± 3.0%). Statistically significant canopy cover differences were found between populations (p = 0.003). The Baldwin County populations was found to have statistically significant different average canopy cover when compared to both the

Geneva and Okaloosa populations (p = 0.008 and p = 0.018, respectively). No statistically significant relationships were found between the number of F. milleri stems and the various microhabitat factors tested.

Soil Sampling

The average pH of all soil samples was acidic (4.860 ± 0.081, Table 4.2). The average pH of 0-10 cm samples was more acidic than the 10-30 cm samples (4.666 ± 0.101 and 5.054 ±

0.109, respectively). The Geneva population had the highest soil pH across both depths (5.454 ±

0.253). The Baldwin population had the second-highest pH (4.834 ± 0.055). The Okaloosa population had the lowest soil pH across both depths (4.668 ± 0.184). Statistically significant differences were found between populations with respect to pH (p = 0.007). The Geneva population was found to have statistically significant differences in pH when compared to both the Baldwin and Okaloosa populations (p = 0.006 and p = 0.017, respectively). The Okaloosa population had the highest C:N ratio (35.3 ± 4.6), however, it was only significantly different from the Baldwin population (p = 0.030). No other significant differences were detected with respect to soil C:N ratio. No statistically significant relationships were found between the number of F. milleri stems and the various soil factors tested.

Soil texture was determined to be sand or sandy loam. The soil texture for all samples within the Geneva County population was sand. Similarly, the majority of soil samples within the Okaloosa County population were also identified to be sand. However, two samples were 68

considered sandy loam and one was identified as loam. The soil texture of the Baldwin population was identified as mostly a sandy loam, but four samples were characterized as loamy sand and two were characterized as a sandy clay loam.

Results from the Web Soil Survey were consistent with the results of the soil texture determined by feel (Thien 1979, Soil Survey Staff 2021). Soils were primarily fine sandy loam or loamy fine sand that is well-drained to somewhat excessively well-drained. Soils were also classified as acidic by the web soil survey, consistent with the findings of the soil samples.

Additionally, both the Baldwin and Okaloosa population had various muck soils that are poorly drained. The National Wetlands Inventory provided similar results. All populations of F. milleri were not directly within a wetland. However, all populations were within approximately 200 m of a freshwater forested or shrub wetland (U. S. Fish and Wildlife Service 2020).

Habitat Suitability Model (HSM)

The preliminary habitat suitability model had an average omission rate of 37.5% and an average area under the curve (AUC) of 0.6 (Figure 4.5). The area covered within the top bin

(0.9-1) of habitat suitability accounted for only 0.5% of the total study area (625 km2). The coastal regions of Alabama and Florida showed higher habitat suitability than inland regions.

The mean temperature of the warmest quarter (bio 10) contributed the most to the overall model

(30.8%). Percent sand within the soil and precipitation in the warmest quarter (bio 18) contributed the second and third most, respectively (27.5% and 20.8%, Table 4.3).

69

Discussion

Environmental stochasticity is the leading cause of extinction for plant species with small, fragmented populations (Matthies et al. 2004). Through personal communication with the local species expert, Ron Miller, it was determined that the three sampled populations were the three largest and most healthy appearing populations. Through our counts, 3,060 stems were found across all three populations. Of the 3,060 stems, 45% (1,367 stems) are found in single patch within the Baldwin County site. This patch is less than 0.5 ha in area. One stochastic event could reduce the total population of F. milleri by 45%. Further, the Baldwin population alone accounts for 79% of the total number of stems found in the three healthiest populations. All patches of F. milleri within the Baldwin population are contained within a 4.5 ha area. Once again, the large proportion of F. milleri stems in a small area makes the species particularly vulnerable to stochastic events. Lastly, all three of the sampled populations, which represent the three largest populations, cover approximately 8.3 ha. The relatively low number of stems found and the small area in which they occupy raises concerns for the perpetuation of the species on the landscape.

Knapp et al. (2021) found 64% of extinct plants in the continental United States and

Canada were single-site endemics. They defined a single site endemic as having an area of occupancy ≤ 6 km2. While F. milleri would not currently be considered a single-site endemic according to their criteria, since it has more than one documented population, it could become a single-site endemic soon if active conservation and proper management are not prioritized.

Knapp et al. (2021) also made the point that preventing a species extinction is a relatively low bar in plant conservation. If F. milleri is to perpetuate on the landscape, simply preventing extinction cannot be the goal. Multiple, self-sustaining populations are needed. For populations

70

to become self-sustaining both sexual and vegetative reproduction will need to take place and genetic variability is needed to enhance resilience. The number of genetically different individuals is likely to be much lower than that of the number of stems counted. Low genetic diversity could hinder the conservation of the species (Spielman et al. 2004). Further, an understanding of the genetic diversity among these 3,060 stems is needed to better inform and justify the need for conservation.

Fothergilla milleri produce seed pods dispersed abiotically by gravity with no known or insect vectors for seed dispersal. Abiotic distribution via gravity may limit the ability for the species to spread and not allow the species to form new populations. Moreover, seed pods found throughout the survey period appeared to be infected by an insect gall (Figure 4.6). No healthy, uninfected seed pods were found in any of the sampled populations. Currently, there is no record of F. milleri fruit being infected in the scientific literature. Although, a post made in

2012 on the online forum BugGuide.net reports a morphologically similar gall on a Fothergilla gardenii in Okaloosa County, Florida (BugGuide 2021). This individual would likely now be considered F. milleri as opposed to F. gardenii. Based on the morphological similarity and species association, this is likely the same species of aphid found in the stem count surveys.

Seeds of American witchhazel (Hamamelis virginiana L.), which is also a member of the

Hamamelidaceae, have been shown to become infected with the aphid Pseudanthonomus hamamelidis (Dietz). In some cases, 76% of seed crops were infected (Steven 1981). F. milleri can reproduce vegetatively, however, a lack of sexual reproduction could severely hamper regeneration and conservation efforts for the species. Further entomological research is needed to identify the species of the aphid and potential avenues for protecting F. milleri from the parasitism will need to be explored.

71

Fothergilla milleri seems to grow in transitional zones between upland conifer forest dominated by loblolly pine and longleaf pine and wetland forests containing Atlantic white- cedar, water oak (Quercus nigra L.), and bald-cypress (Taxodium distichum (L.) Rich.). An analysis of the populations within the National Wetlands Inventory further supports this conclusion. All patches of F. milleri were within 200 m of a designated wetland, many were close to or on the delineation line (U.S. Fish and Wildlife Service 2020). However, none were within the wetlands themselves. These unique transitional habitats could face a combination of threats from both the uplands and wetlands. Many of the upland conifer stands adjacent to F. milleri patches are managed pine plantations. Intensive forest management, such as midstory and ground cover herbicide application and harvesting operations, could negatively impact these F. milleri populations.

Many patches of F. milleri are under a thick midstory of large gallberry and buckwheat tree. If active management is not carried out, F. milleri populations may become heavily shaded and may eventually be out competed for both light and space. Additionally, densiometer measurements show relatively high canopy cover. Haynes et al. (2020) discuss the need for regular fire to constrain competing vegetation in F. gardenii populations. Additionally, fire suppression has been cited as one of the major factors leading to habitat degradation in F. gardenii populations, some of which may now be considered F. milleri (Barger et al. 2013). Fire is likely needed to impede buckwheat tree and large gallberry within F. milleri populations to prevent the populations from becoming shaded out. Further research is needed into the role fire may potentially play in F. milleri regeneration to better inform conservation land management for the species.

72

The results of this study support the habitat description Haynes et al. (2020) provides when first describing the species. However, this study provides a more in-depth analysis of the habitat that supports F. milleri. Haynes et al. (2020) describe the habitat of F. milleri as, “sandy peat shrub bogs, seepages, dry longleaf pine woodlands, and the edge of Cyrilla racemiflora or

Taxodium ascendens swamp forest”. Although no T. ascendens was found within the study plots, these habitat descriptions align with the results of this study. Additionally, Barger et al. (2013) describes the habitat of F. gardenii within Baldwin County, Alabama, the population described by Barger et al. (2013) is likely the same Baldwin County population sampled for this study and would likely now be considered F. milleri. Barger et al. (2013) describe the habitat as, “wet savannas, flatwoods, pocosins, and along the edges of pitcherplant bogs, baygalls, and shrub swamps”. The results of this study once again support this description of the species’ habitat.

Particularly, the edge or transitional habitat between the flatwoods and bogs.

Fothergilla milleri and F. gardenii seem to share similar habitat requirements. The habitat of F. gardenii is described to be pine savannas and evergreen bogs, like that of F. milleri

(Barger et al.2013). Further, F. gardenii is a noted species associate of Atlantic white-cedar, which was also the third most abundant overstory species and fourth most abundant midstory species within the F. milleri vegetation plots (Sheridan et al. 2003).

The preliminary habitat suitability model presented in this study further supports the conclusion that F. milleri fills a unique niche in coastal forests. Only 625 km2, or 0.5% of the total study area, were found to be in the highest bin of habitat suitability. The findings of this preliminary habitat suitability model aligned with the results from the soil sampling within populations of F. milleri. Percent sand in the soil contributed 27.5% to the overall model and was the second most important environmental variable. These results show soil type, particularly the

73

sand component, is consistent across populations and potentially important for F. milleri establishment and perpetuation. The study area was intentionally minimized to avoid extrapolation beyond the bounds of the model. With the limited number of occurrence points available for use in the habitat suitably model, there was concern straying too far from documented occurrences could result in inappropriate conclusions. Small numbers of occurrence records can be problematic when developing such models. However, species with few occurrence records are often the species most in need of habitat suitability modeling to inform accelerated conservation action before more occurrences can be found (Gaubert et al. 2006). Due to its more recent discovery, there are no reliable records of F. milleri in herbariums or within citizen science databases. Future habitat suitability models may benefit in time as more F. milleri records becomes established in both formal herbariums and citizen science databases.

The default parameters setting within Maxent have been shown to overfit models and in some situations and do not produce the best possible models (Phillips et al. 2006, Anderson and

Gonzalez 2011, Radosavljevis and Anderson 2014). A systematic review of the models utilizing

Maxent found 3.7% of studies evaluated both the regularization multiplier and the feature class as this study did (Morales et al. 2017). This habitat suitability model provides the first attempt at directing conservation efforts for F. milleri, including possible searches for new populations.

Further, this model could inform conservation action for F. milleri by prioritizing areas of high habitat suitability for introduction or reintroduction.

The evidence suggests F. milleri is facing several threats including, environmental stochasticity, lack of sexual reproduction, and heavy competition. Due to the small population size and increasing threat, F. milleri may be a candidate for listing through the Endangered

Species Act. Further evaluation is needed to determine F. milleri’s eligibility to be federally

74

listed. Nevertheless, research into both the role fire plays in the regeneration of the species, the species ability to sexually reproduce, and genetic diversity and gene flow are all needed to fully understand the current conservation status of F. milleri. Concurrently, direct action to protect F. milleri is needed. Possible conservation measures include ex-situ conservation in botanic gardens, cultivating the species to supplement current populations, establishing seed banks and growing in plant nurseries, and further protection of wild populations. Lastly, within wild populations, land management such as midstory and overstory reduction cuts or prescribed fire may be needed to prevent populations from becoming outcompeted for limited resources such as light and space.

Conclusions

Fothergilla milleri occupies a unique habitat in the transition between upland conifer forest and wetlands. Further, F. milleri likely requires specific soils that are acidic and have high sand content. Fothergilla milleri is facing multiple threats to its perpetuation. The most pressing being environmental stochasticity. One such event could eliminate nearly 79% of the stems within the three healthiest populations. Further, it appears sexual reproduction is not currently taking place in the sampled populations of F. milleri due to the fruit being infected by an unknown aphid. Also, due to the small population size and disjunct nature of those populations, a lack of genetic diversity is a concern that needs further research. Lastly, many patches of F. milleri are under thick midstories. Fire may be needed to open the midstory so F. milleri is not shaded out in these patches. A lack of suitable habitat may further hinder conservation efforts.

Based on the identified threats and a very specific habitat of F. milleri, it is likely active and intentional conservation efforts are needed for F. milleri to remain on the landscape in perpetuity. 75

Acknowledgments

We would like to thank Dr. Ali Paulson and Dr. Carlos Ramirez- Reyes for their assistance in developing the habitat suitability model. This work would not have been possible without the help, flexibility, and accommodations provided by Dr. Wayne Barger and the land managers of the Perdido Wildlife Management Area. We would also like to thank Colin Davies of American Forest Management and Tyler Macmillan of the Northwest Florida Water

Management District for securing permission to access F. milleri populations. This work is the combination of the Forest and Wildlife Research Center and the Mississippi Agriculture and

Forest Experiment Station, Mississippi State University, and this material is based upon work supported by the U.S. Department of Agriculture, National Institute for Food and Agriculture,

McIntire-Stennis project under accession number 1019117. Any opinions, findings, conclusions, or recommendations are those of the authors and do not necessarily reflect the views of the U.S.

Department of Agriculture.

76

Tables

Table 4.1 Associated woody plant species within Miller’s witch-alder populations in Geneva County (Alabama), Okaloosa County (Florida), and Baldwin County (Alabama)

Species Stems/ha Basal area/ha (m2) Overstory Pinus taeda 161 (57) 5.87 (1.96) Pinus palustris 80 (36) 4.47 (2.09) Chamaecyparis thyoides 79 (41) 2.37 (1.18) Cliftonia monophylla 48 (20) 0.59 (0.26) Pinus elliottii 41 (19) 2.95 (1.16) Magnolia virginiana 19 (9) 0.35 (0.14) Liriodendron tulipifera 11 (6) 0.41 (0.23) Quercus nigra 11 (5) 0.45 (0.26) Quercus virginiana 9 (8) 0.30 (0.28) Nyssa sylvatica 8 (4) 0.29 (0.21) Quercus phellos 6 (4) 0.09 (0.06) Ilex opaca 4 (3) 0.06 (0.04) Taxodium distichum 4 (3) 0.20 (0.15) Acer rubrum 3 (3) 0.03 (0.03) Cyrilla racemiflora 1 (1) 0.02 (0.02) Midstory Cliftonia monophylla 213 (88) 0.54 (0.26) Ilex coriacea 101 (64) 0.11 (0.07) Magnolia virginiana 64 (28) 0.09 (0.03) Chamaecyparis thyoides 43 (23) 0.16 (0.10) Cyrilla racemiflora 37 (27) 0.04 (0.03) Acer rubrum 27 (11) 0.05 (0.02) Nyssa sylvatica 24 (10) 0.07 (0.03) Pinus taeda 24 (15) 0.09 (0.05) Quercus nigra 13 (5) 0.02 (0.01) Quercus phellos 13 (13) 0.01 (0.01) Liriodendron tulipifera 5 (5) 0.02 (0.02) Myrica cerifera 5 (5) 0.01 (0.01) Pinus palustris 5 (5) 0.02 (0.02) Taxodium distichum 5 (4) 0.02 (0.01) Ilex vomitoria 3 (3) 0.00 (0.00)

Average stems/ha and basal area/ha of associated woody plant species within the three sampled populations of Fothergilla milleri (W. D. Phillips & J. E. Haynes, Hamamelidaceae). Standard errors are presented parenthetically. 77

Table 4.2 Soil characteristics of the three sampled Miller’s witch-alder populations in Geneva County (Alabama), Okaloosa County (Florida), and Baldwin County (Alabama)

C:N Population Depth Texture pH Ratio All All Sand - Sandy Clay Loam 4.9 (0.1) 26.9 (1.8) All 0-10 Sand - Sandy Clay Loam 4.7 (0.1) 27.8 (2.8) All 10-30 Sand - Sandy Clay Loam 5.1 (0.1) 26.0 (2.4) Geneva All Sand 5.5 (0.3) 21.3 (3.1) Geneva 0-10 Sand 5.0 (0.2) 22.8 (3.5) Geneva 10-30 Sand 5.9 (0.1) 19.8 (6.4) Okaloosa All Sand - Sandy Loam 4.7 (0.2) 35.3 (4.6) Okaloosa 0-10 Sand - Sandy Loam 4.3 (0.2) 37.6 (7.2) Okaloosa 10-30 Sand - Sandy Loam 5.0 (0.2) 33.1 (6.3) Baldwin All Sandy Loam - Loamy Sand - Sandy Clay Loam 4.8 (0.1) 23.5 (1.0) Baldwin 0-10 Sandy Loam - Loamy Sand - Sandy Clay Loam 4.8 (0.1) 23.5 (1.6) Baldwin 10-30 Sandy Loam - Loamy Sand - Sandy Clay Loam 4.9 (0.1) 23.4 (1.3)

Soil characteristics of the three sampled populations of Fothergilla milleri (W. D. Phillips & J. E. Haynes, Hamamelidaceae). Standard errors are presented parenthetically.

78

Table 4.3 Contribution of environmental variables to Miller’s witch-alder habitat suitability model

Environmental Variable Contribution (%) Mean Temperature of Warmest Quarter (bio10) 30.8 Percent Soil Sand Content 27.5 Precipitation in Warmest Quarter (bio18) 20.8 Distance to Water 11.3 Isothermality (bio3) 6.7 Mean Temperature of Coldest Quarter (bio11) 1.3 Annual Mean Temperature (bio1) 0.8 Soil pH 0.5 Precipitation Seasonality (bio15) 0.2 Max. Temperature of Warmest Month (bio5) 0.0 Precipitation of Wettest Quarter (bio16) 0.0 Precipitation of the Driest Quarter (bio17) 0.0 Precipitation of Driest Month (bio14) 0.0 Precipitation of Wettest Month (bio13) 0.0 Annual Precipitation (bio12) 0.0 Mean Diurnal Range (bio2) 0.0 Precipitation of Coldest Quarter (bio 19) 0.0 Temperature Seasonality (bio4) 0.0 Min. Temperature of Coldest Month 0.0 Temperature Annual Range (bio7) 0.0

Relative contribution to of the 20 environmental variables used in developing the preliminary habitat suitability model for Fothergilla milleri (W. D. Phillips & J. E. Haynes, Hamamelidaceae).

79

Figures

Figure 4.1 Reproductive and vegetative morphology of Miller’s witch-alder in Baldwin County, Alabama, United States

(a) Typical reproductive morphology of Fothergilla milleri (W. D. Phillips & J. E. Haynes, Hamamelidaceae) (b) Typical vegetative morphology of Fothergilla milleri. Photos taken by the authors.

80

Figure 4.2 Locations of Miller’s witch-alder populations sampled in Alabama and Florida, United States

The location of the three populations of Fothergilla milleri (W. D. Phillips & J. E. Haynes, Hamamelidaceae) sampled. These populations were determined to be the healthiest through personal communication with local species expert, Ron Miller of Pensacola, Florida.

81

Figure 4.3 Plot diagram for sampling of Miller’s witch-alder habitat

Plot layout for sampling overstory stems, midstory stems, canopy cover, and soil of the three sampled Fothergilla milleri (W. D. Phillips & J. E. Haynes, Hamamelidaceae) populations. Arrows indicate directionality of canopy cover measurement.

82

Figure 4.4 Stem counts of Miller’s witch-alder populations in Baldwin County (Alabama), Okaloosa County (Florida), and Geneva County (Alabama)

Stem counts of Fothergilla milleri (W. D. Phillips & J. E. Haynes, Hamamelidaceae) for each of the three populations sampled. Shading indicates different F. milleri patches within each population.

83

Figure 4.5 Habitat suitability model developed for Miller’s witch-alder in the southeastern United States

The habitat suitability model developed within Maxent for Fothergilla milleri (W. D. Phillips & J. E. Haynes, Hamamelidaceae) from historic and extant occurrence points (n = 8) and twenty environmental variables. The study area in indicated by the grey shading. The model had an overall average omission rate of 37.5% and an AUC of 0.6.

84

Figure 4.6 Healthy and infected seed capsules of Miller’s witch-alder Baldwin County, Alabama, United States

(a) The seed capsules of a healthy Fothergilla milleri (W. D. Phillips & J. E. Haynes, Hamamelidaceae) (b) An infected seed capsule of F. milleri by and unknown aphid. Photos taken by the authors.

85

Literature Cited

Abdelaal, M., M. Fois, G. Fenu, and G. Bacchetta. 2019. Using MaxEnt modeling to predict the potential distribution of the endemic plant Rosa arabica Crép. in Egypt. Ecological Infomatics 50:68-75.

Anderson, R.P., and I. Gonzalez Jr. 2011. Species-specific tuning increases robustness to sampling bias in models of species distributions: an implementation with Maxent. Ecological Modeling 222:2796-2811.

Barger, T.W., D.D. Spaulding, and B.D. Holt. 2013. The vascular flora of the Perdido River Forever Wild tract, Baldwin County, Alabama. Castanea 78(2):119-133.

Brown J.L., J.R. Bennett, and C.M. French. 2017. SDMtoolbox 2.0: The next generation Python- based GIS toolkit for landscape genetic, biogeographic and species distribution model analyses. Peerj 5:e4095.

BugGuide. 2021. Gall and aphids in fruit of very rare Fothergilla gardenii (Hamamelidaceae). Available at < https://bugguide.net/node/view/618940>. Accessed 26 March 2021.

Costa, G.C., C. Nogueira, R.B. Machado, and G.R. Colli. 2010. Sampling bias and the use of ecological niche modeling in conservation planning: a field evaluation in a biodiversity hotspot. Biodiversity and Conservation 19:883-899.

Cottam, G. and J.T. Curtis. 1956. The use of distance measures in phytosociological sampling. Ecology 37(3):451-460.

Fick, S.E. and R.J. Hijmans. 2017. WordClim 2: New 1km spatial resolution climate surfaces for global land areas. International Journal of Climatology 37(12):4302-4315.

Gary, R.H., Z.D. Wilson, C.M. Archuleta, F.E. Thompson, and J. Vrabel. 2009. Production of a 1:1,000,000-scale hydrography dataset for the United States-feature selection, simplification, and refinement. U.S. Geological Survey Scientific Investigations Report. 2009-5202. 22.

Gaubert, P., M. Papeş, and A.T. Peterson. 2006. Natural history collections and the conservation of poorly known taxa: Ecological niche modeling in central African rainforest genets (Genetta spp.). Biological Conservation 130(1):160-117.

Haynes, J.E., W.D. Phillips, A. Krings, N.P. Lynch, and T.G. Ranney. 2020. Revision of Fothergilla (Hamamelidaceae), including resurrection of F. parvifolia and a new species, F. milleri. PhytoKeys 144:57-80.

Hemati, T., S. Pourebrahim, M. Monavari, and A. Baghvand. 2020. Species-specific nature conservation prioritization (a combination of maxent, co$sting nature, and DINAMICA EGO modeling approaches). Ecological Modeling 429. 86

Hernandez, P.A., C.H. Graham, L.L. Master, and D.L. Albert. 2006. The effect of sample size and species characteristics on performance of different species distribution modeling methods. Ecography 29(5):773-785.

Knapp, W.M, A. Frances, R. Noss, R.F.C. Naczi, A. Weakley, G.D. Gann, B.G. Baldwin, J. Miller, P. McIntyre, B.D. Mishler, G. Moore, R.G. Olmstead. A. Strong, K. Kennedy, B. Heidel, and D. Gluesenkamp. 2021. Vascular plant extinction in the continental UnitedStates and Canada. Conservation Biology 35(1):360-368.

Kumar, S. and T.J. Stohlgren. 2009. Maxent modeling for predicting suitable habitat for threatened and endangered tree Canacomyrica monticola in New Caledonia, Journal of Ecology and Natural Environment 1(4):94-98.

Matthies, D., I. Bräuer, W. Maibom, and T. Tscharntke. 2004. Population size and the risk of local extinction: empirical evidence from rare plants. Oikos 105(3):481-488.

Menges, E.S. 1992. Stochastic modeling of extinction in plant populations. Pp. 253-275 in P.L. Fiedler and S.K. Jain, eds., Conservation Biology. Springer. Boston, MA.

Merow, C., M.J. Smith, and J.A. Silander Jr. 2013. A practical guide to maxent for modeling species’ distributions: what it does, and why inputs and settings matter. Ecography 36:1058-1069.

Morales, N.S., I.C. Fernández, and V. Baca-González. 2017. Maxent’s parameter configuration and small samples: are we paying attention to recommendations? A systematic review. PeerJ 5:e3093.

NatureServe Explorer. 2021. An Online Encyclopedia of Life: Fothergilla milleri. Available at . Accessed 29 March 2021.

Pearson, R.G., C.J. Raxworthy, M. Nakamura, and A.T. Peterson. 2007. Predicting species distributions from small numbers of occurrence records: a test case using cryptic geckos in Madagascar. Journal of Biogeography 34(1):102-117.

Phillips, S.J., R.P. Anderson, and R.E. Schapire. 2006. Maximum entropy modeling species geographic distributions. Ecological Modeling 190:231-259.

Phillips S.J. and M. Dudík. 2008. Modeling of species distributions with MaxEnt: New extensions and a comprehensive evaluation. Ecography 31:161-175.

R Core Team. 2020. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. < https://www.r-project.org/> Accessed 22 August 2021.

87

Radosavljevic, A., and R.P. Anderson. 2014. Making better maxent models of species distributions: complexity, overfitting, and evaluation. Journal of Biogeography 41:629- 643.

Ranney, T.G., N.P Lynch, P.R. Fantz. and P. Cappiello. 2007. Clarifying and Nomenclature of Fothergilla (Hamamelidaceae) Cultivars and Hybrids. HortScience 42(3):470-473.

Remirez-Reyes, C., G. Street, F.J. Vilella, D.T. Jones-Farrand, M.S. Wiggers, and K.O. Evans. 2021. Ensemble Species Distribution Model Identifies Survey Opportunities for At-Risk Bearded Beaksedge (Rhynchospora crinipes) in the Southeastern United States. Natural Areas Journal 41(1):55-63.

Remya, K., A. Ramachandran, and S. Jayakumar. 2015. Predicting the current and future suitable habitat distribution of Myristica dactyloides Gaertn. using MaxEnt model in the Eastern Ghats, India. Ecological Engineering 82:184-188.

Sharma, S., K. Arunachalam, D. Bhavsar, and R. Kala. 2018. Modeling habitat suitability of Perilla frutescens with maxent in Uttarakhand-A conservation approach. Journal of Applied Research on Medicinal and Aromatic Plants 10:99-105.

Shcheglovitova, M. and R.P. Anderson. 2013. Estimating the optimal complexity for ecological niche models: A jackknife approach for species with small sample sizes. Ecological Modeling 269:9-17.

Sheridan, P.M., and Patrick, T.S. 2003. A rare plant survey of Atlantic white-cedar habitats of western Georgia. Pp. 101-112. In Proceedings of the Atlantic White-cedar Restoration Ecology and Management Symposium, 31 May - 2 June 2000, Newport News, Virginia.

Soil Survey Staff. 2019. Soil Survey Geographic (SSURGO) Database for the United States. Natural Resources Conservation Service, U.S. Department of Agriculture, Washington, D.C. Accessed 22 August 2021.

Soil Survey Staff. 2021. Web Soil Survey. Natural Resources Conservation Service, U.S. Department of Agriculture, Washington, D.C. Accessed 22 August 2021.

Spielman, D., B.W. Brook, and R. Frankham. 2004. Most species are not driven to extinction before genetic factors impact them. Proceedings of the National Academy of the Sciences 101(42):15261-15264.

88

Stevens, D.D. 1981. Abundance and survival of a seed-infesting weevil, Pseudanthonomus hamamelidis (Coleoptera: ), on its variable-fruiting host plant, witch-hazel (Hamamelis virginiana). Ecological Entomology 6:387-396.

Syfert, M.M., M.J. Smith, and D.A. Coomes. 2013. The effects of sampling bias and model complexity on the predictive performance of Maxent species distribution models. PLoS One 8(2):e55158.

Thien, S.J. 1979. A flow diagram for teaching texture-by-feel analysis. Journal of Agronomic Education 8(1):54-55.

U. S. Fish and Wildlife Service. 2020. National wetlands inventory website. Fish and Wildlife Service, U.S. Department of the Interior, Washington, D.C. Accessed 22 August 2021

U.S. Geological Survey. 2010. National watershed boundary dataset (WBD). U.S. Department of the Interior, Washington, D.C. < https://www.usgs.gov/core-science- systems/ngp/national-hydrography/watershed-boundary-dataset?qt- science_support_page_related_con=4#qt-science_support_page_related_con> Accessed 22 August 2021

Wisz, M.S., R.J. Hijmans, J. Li, A.T. Peterson, C.H. Graham, A. Guisan, and NCEAS Predicting Species Distributions Working Group. 2008. Effects of sample size on the performance of species distribution models. Diversity and Distributions 14(5):763-773.

Yang, X.Q., S.P.S. Kushwaha, S. Saran, J. Xu, and P.S. Roy. 2013. Maxent modeling for predicting the potential distribution of medicinal plant, Justicia adhatoda L. in Lesser Himalayan foothills. Ecological Engineering 51:83-87.

89

CHAPTER V

CONCLUSION

As rare and imperiled plant species face increased threats to their perpetuation direct and intentional action is needed. Various governmental and non-governmental agencies, such as botanical gardens and state plant conservation alliances are working to save at-risk plants from extinction, however, more is needed. Recently, “The Great USA Plant Conservation Challenge” was published in the journal Biodiversity and Conservation (Molano-Flores 2021). Within this document, there are three challenges, first for foundations to fund research efforts for all federally listed plant taxa. As of 2011, plants made up 57% of federally endangered species, yet they received less than 4% of funding (Havens et al. 2014). For plant conservation to succeed, simply put, more funding from government agencies, non-government agencies, and academia is needed. The second challenged posed within the “Great USA Plant Conservation Challenge” is for academic institutions to sponsor research on all federally listed plant taxa. This will take closer coordination and collaboration from various institutes of higher learning. In the

Southeastern United States state plant conservation alliances are being formed to meet this need, the first of which was founded in Georgia in 1995. More recently, the Southeastern Plant

Conservation Alliance was formed in March of 2020. These alliances link various academic, governmental agencies, botanical gardens, and other non-governmental organizations to more effectively direct limited plant conservation funding and manpower. As more aspiring scientists become interested in plant conservation there should be the manpower, if there is the motivation

90

to save many rare and imperiled plant species. The last challenge is for all scientists, including those that do not explicitly work in plant conservation, to pay more attention to vulnerable plant taxa. Scientists sympathetic to plant conservation should start to ask themselves how their research could be utilized for plant conservation. For example, a restoration ecologist could collaborate with a plant conservationist to plant a native imperiled species rather than mass produce seed mixes. The three challenges laid out in the “Great USA Plant Conservation

Challenge” are clear necessary steps needed if rare and imperiled plants are to be protected.

Within this thesis, we met several of the challenges of the “Great USA Plant

Conservation Challenge”. Per the first challenge, funding, in the form of a departmental assistantship, was directed towards plant conservation research. However, this thesis primarily addressed the second challenge, for academia to sponsor research on rare and imperiled plant species. The three species studied in this thesis do not tend to receive conservation attention. Yet, all are considered imperiled either at the state or global level. Through this research, we now have a better ecological understanding of three imperiled plant species. Allowing us to provide more appropriate conservation management recommendations. Through researching these species, we collaborated with various scientist that does not focus strictly on plant conservation to better understand and manage these species, addressing the third challenge. This included, but was not limited to, a hydrologist, an ichthyologist, and a statistician, all of which contributed their specific expertise’s to better understand these rare and imperiled species. Additionally, we collaborated with various academic, governmental, and non-governmental agencies to conduct this research. Beyond the greater understanding of the three imperiled species studied, this thesis provides a first attempt and potential road map for other academic institutions to attempt to meet the difficult challenges laid out in “The Great USA Plant Conservation Challenge”. Further, this

91

thesis can help facilitate additional research and conservation for the three species studied and other rare and imperiled plant species.

Overall, this thesis contributes to the ecological understanding of three species, mountain stewartia, Atlantic white-cedar, and Miller’s witch-alder, that can therefore contribute to their conservation. Utilizing cutting edge technology, land can now be identified for both mountain stewartia and Miller’s witch-alder to be protected or for populations to be established. Further, understanding the regeneration potential of Atlantic white-cedar post- tropical cyclone disturbance allows land managers and conservatist the ability to better manage for the species.

Lastly, numerous threats to the perpetuation of Miller’s witch-alder, such as pest, lack of fire, and environmental stochasticity, have now been identified to assess conservation priorities for the species.

92

Literature Cited

Havens, K., A.T. Kramer, and E.O. Guerrant Jr. 2014. Getting plant conservation right (or not): The case of the United States. International Journal on Plant Sciences 175(1):3-10.

Molano-Flores, B. 2021. The great USA plant conservation challenge. Biodiversity and Conservation 30:1595-1597.

93