I LLINOI S UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN

PRODUCTION NOTE

University of Illinois at Urbana-Champaign Library Large-scale Digitization Project, 2007.

The Population Biology of Torreya taxifolia: Habitat Evaluation, Fire Ecology, and Genetic Variability

Mark W. Schwartz and Sharon M. Hermann

Center for Biodiversity Technical Report 1992(Z)

Illinois Natural History Survey 607 E. Peabody Drive Champaign, Illinois 61820

Tall Timbers, Inc. Route 1, Box 678 Tallahassee, Florida 32312

Prepared for Florida Game and Freshwater Fish Commission Nongame Wildlife Section 620 S. Meridian Street Tallahassee, Florida 32399-1600

Project Completion Report NG89-030

TABLE OF CONTENTS page Chapter 1: Species background and hypotheses for...... 5 the decline of Torreya taxifolia,

species Background ...... 6 Hypotheses for the Decline...... 0 Changes in the Biotic Environment ...... 10 Changes in the Abiotic Environment ..... 13

Discu~ssion *0o ** eg. *.*. 0 0*.0.*09 6 0 o**** o*...21

Chapter 2: The continuing decline of Torreyap iola....2 Study.Area and Methods ooo...... 25

Results * ** ** ** ** ** ** ...... 30

Chapter 3: Genetic variability in Torreya taxif-olia...... 4

Methods...... * 0 C o490 0

Results ...... *oe*...... o51

-0L-icmion *.. ~ 0000 00000@55 Management _Recommendations .000000000000.0.60

Chapter 4: The light relations of Tgr .taz'ifgli with ..... 62 special emphasis on the relationship to growth and,,disease- Methods o...... 0.0.0.0.0.00.eoo63

Light and Growth ...... 64 Measurements'-of photosynthetic rates 0,.65

Light and Growth ...... 69 Measurements of photosynthetic rates ..71.

Discussion...... *0* * * * * * * ** . 81

Chapter 5: The foliar fungal associates of Torreya...... 85 ta ifola: pathogenicity and susceptibility to smoke Methods 0 0 0... 0..0.....0..0..0..0..0..0..0..0..0. *86 Fungal Isolation 00000-0900960000000690086 Pathogenicity Tests *00000....00.0.0.087 Fungal.Susceptibility to Smoke ...... 87 Field Tests of Smoke Hypothesis ...... 90 Results...... 00000000000000 .000000000*91 Fungal Isolation and Pathogenicity ....91, Smoke Susceptibility ..... oo...... 92 Discussion...... 000 00000000000098

2 FIGURES

Chapter 1 page 1.1 Range Map of Torreya taxifolia ...... 7 1.2 Annual Precipitation in Blountstown, FL, 1922-1982 ...... 14 1.3 Water Temperature Data, Above and Below Lake Seminole ...16 1.4 Air Temperature Data, Blountstown, FL, 1931-1984 ...... 18 Chapter 2 2.1 Range Map Detailing Study Areas ...... 26 2.2 Stem Size Distribution for Census Population ...... 31. 2.3 Number of Stems per Individual ...... 31 2.4 Distribution of T. taxifolia With Respect to Elevation ..34 2.5 Distribution With Respect to Canopy Cover ...... 34 2.6 Distribution With Respect to Aspect ...... 34 2.7 Apparent Age Distribution of Census Population ...... 36 2.8 Age Correction Regression ...... 36 Chapter 3 3.1 Map of Sampled Populations...... 50 BLANK PAGE ...... *****..* ***************. * * *.. 53 3.2 Cluster Diagram of Populations ...... 58 Chapter 4 4.1 Comparison of Growth by Light ...... 70 4.2 Light Saturation Photosynthesis Curve ...... 73 4.3 Photosynthetic Rates of Diseased Versus Healthy Needles .74 4.4 Photosynthetic Rates by Needle Age ...... 75 4.5 Light Saturation Curves for Five Closely Related Species.77 4.6 Photosynthetic Rates Across Relative Humidity, 5 Species.78 4.7 Photosynthetic Rates Across Temperature, 5 Species ...... 79 4.8 Rate of Water Loss in Absence of Watering, 5 Species ....80 4.9 Photosynthetic Rates Across.Water Stress, 5 Species ..... 80 Chapter 5 5.1 Smoke Residue Deposition Versus Time Treatment ...... 89 5.2 Effects of Direct Smoke Treatment on Fungal Growth ...... 93 5.3 Effects of Smoke Residue on Fungal Growth ...... 94 5.4 Effects of Smoke Residue on Spore Germination Rate ...... 95 5.5 Effects of Smoke Residue on Spore concentration ...... 96 5.6 Effects of Smoke Residue on Spore'Growth Rate ...... 97

Chapter 6 None TABLES

Chapter 1 page 1.1 Fungal Associates of Torreya taxifolia ...... 9 1.2 Hypotheses for the Decline of T. taxifolia ...... ll Chapter 2 2.1 Census Population ...... 28 2.2 Mean Length of Longest Stem, by Population ...... 33 2.3 Frequency of Disease Symptoms ...... 38 2.4 Growth Fate of T. taxifolia, by Size Class ...... 38 2.5 Growth Fate of T. taxifolia, by Disease Load ...... 42 Chapter 3 3.1 Allele Frequencies for Enzymes Studied ...... 52 3.2 Heterogeneity of Sampled Populations ...... 54 3.3 Genetic Distance Between Sampled Populations ...... 56 3.4 Genetic Diversity Within and Between Populations ...... 57 Chapter 4 4.1 Growth Fate of Trees in Different Canopy Cover Classes .72 4.2 Disease Incidence of Trees in Canopy Cover Classes ..... 72

Chapter 5 5.1 Fungi Isolated From T. taxifolia Used in Experiments ...89

Chapter 6 None CHAPTER 1 Species background and hypotheses for the decline of Torreva taxifolia

During the late 1950's, Torreya taxifolia suffered from a catastrophic decline, presumably fungal in origin, that decimated adult populations throughout its range (Godfrey and Kurz 1962). The cause of this decline remains a mystery; no specific pathogen has been implicated (Alfieri et al. 1967, USFWS 1986); the predominant speculation is that any of a combination of several environmental changes rendered T. taxifolia more susceptible to native pathogens (USFWS 1986). Four hypotheses for the cause of the decline of T. taxifolia have been proposed (Alfieri et al. 1967, Toops 1981, Barnes 1985, USFWS 1986). Correlative evidence to address two of these hypotheses is presented here. In addition, five new hypotheses for the decline, and evidence to assess two of these hypotheses, are presented here. Direct evidence to adequately test any of the hypotheses is mostly lacking because no healthy individuals remain in the wild, no undisturbed habitat remains, and the primary pathogen eludes discovery. With this lack of direct evidence, we are left to piece together anecdotal and correlative evidence as to the cause of the dieback. Providing management to facilitate recovery of T. taxifolia is also hindered by a lack of sufficient information. The purpose

5 of this chapter is to discuss the relative merits of different hypotheses proposed to explain the decline. The results presented here point toward management recommendations despite an inherent inability for adequate testing of virtually any hypothesis. This chapter discusses, by example, the difficulty in isolating pathogen-induced declines in endangered species, and how to approach management recommendations given this information gap.

SPECIES BACKGROUND Florida torreya (T. taxifolia, Taxaceae) is an endemic conifer restricted to ravine slopes along the eastern side of the

Apalachicola River in the central panhandle of Florida (Figure

1.1). This dioecious evergreen tree can grow up to 20m in height at maturity (Godfrey 1988) and was formerly a common understory tree in the region (Chapman 1885, Harper 1914, Reinsmith 1934, Kurz 1838a, 1938b). The ravine habitats which support T. taxifolia are characterized by a diverse mixture of hardwoods and conifers, most notably beech (Fagus grandifolia), and magnolia

(Magnolia grandiflora), along with various oaks and hickories along upper slopes (Platt and Schwartz 1990, Schwartz 1990). Several factors reduced the abundance of the species prior to the decline. In the 1800s, T.. taxifolia was cut along the

Apalachicola River for fuel (Chapman 1885, Gray 1889), and for fenceposts and shingles through the 1900s (Chapman 1885,

Reinsmith 1934, Kurz 1938b, Burke 1978). However, T. axifolia

only became endangered after a catastrophic decline during the 1950s. This decline is thought to have been caused by fungal Figure 2.2.. The range of Torreya taxifolia, including an outlying population in the vicinity of Lake Ocheese-e. Habitat features digitized from 1953 and 1955 USDA aerial photographs using ARC/INFO ver 5.0. Tgrreya~tYa x±i.I is restricted within its range to floodplain and ravine habitats.

Rangeof Torreya taxifolia Suitable Habitat Other Land Cover r -r 1 10 KM *Floodplain *Towns flRavines SAgriculture ElForested Uplands

ILa Uswekemined From USOA kqIW PhonogrrKgy, 1953-105 pathogens, although no primary pathogen has been identified

(Godfrey & Kurz 1962, Alfieri et al. 1967, 1984). The apparent lack of an introduced pest is unusual among catastrophic diebacks of fungal origin (Manion 1981). By 1962, virtually no adult T. taxifolia remained in the wild (Godfrey and Kurz 1962). Recent censuses indicate that approximately 1000 juveniles, and no adults, remain in the wild (Baker 1982, Chapter 2). Most individuals in the wild are less than 2 m tall, are sexually immature, and carry symptoms of foliar diseases (Chapter 2). Few specifics were recorded about the decline of the Tv taxifolia. The earliest observation of disease symptoms may have been in 1938 (Nieland, pers. comm. as cited in Alfieri al.

1967), however, other reports from this decade make no mention of disease (Reinsmith 1934, Kurz 1938a, 1938b). The next report of disease came from Torreya State Park in 1955 (USFWS 1986). By the time the decline was first documented, adult populations had already been decimated (Godfrey.and Kurz 1962). No information is available on the rate of the decline, exact symptoms of disease, or the location of any disease epicenter if one existed.

Symptoms of disease in T. taxifolia at present are needle spots, needle necrosis and stem cankers (Chapter 2). Early reports of disease cited low vigor and needle blight (Godfrey & Kurz 1962). The fungal associates isolated from L. taxifolia do not include introduced obvious pathogens (Table 1.1). The recovery plan for T. taxifolia (USFWS 1986) cites the lack of an introduced pathogen to suggest that environmental changes may Table 1.1. A list of potential fungal associates of Torreya taxifolia, including species isolated from diseased tissue (A), as well as fungi hypothesized to be in association with T. taxifolia (B) (from Alfieri et al. 1967, 1984). A) Fungi isolated from Torreya taxifolia 1) Phyllosticta sp. (needle spot)* 2) Macrophoma sp. (needle blight)* 3) Fusarium sp. (root rot)* 4) Pestalotia natens (needle spot)* 5) Acremonium sp. (?)* 6) Botrosphaeria sp. (needle spot) 7) Xvlocoremium flabelliforme (?)* 8) Sclerotium rolfsii (southern blight) 9) Rhizoctonia solani (root rot) 10) Sphaeropsis sp. (needle blight) 11) Physalospora sp. (twig and needle blight) ? B) Fungi hypothesized to be associated with . taxifolia 12) Alternaria sp. (needle spot) 13) Diplodia natalensis (twig dieback) 14) Pythium sp. (root rot) 15) Phytophthora cinnamomi (root rot) * - tested for pathogenicity with negative results

9 have created plant stress, which in turn may have rendered T. taxifolia more susceptible to infection by native pathogens. The native pathogen hypothesis requires additional framework because these pathogens generally do not reach epidemic proportions unless the host plants are under stress (Schoeneweiss 1978). HYPOTHESES FOR THE DECLINE Four previously proposed hypotheses along with five new hypotheses are presented (Table 1.2). Very little data can be brought to bear directly upon many of these hypotheses. Each of these nine hypotheses is briefly discussed, citing any evidence, upon which to assess the validity of the hypotheses. HYPOTHESES RELATED TQ CHANGES IM THE BIOTIC ENVIRONMENT Al Introduced Fungal Pathogen Exotic pathogens could invade the range of T. taxifolia in any of a number of ways. Perhaps the most likely means for introduction of a pathogen that could have caused die-back in T. taxifolia was through slash pine plantations in uplands adjacent to ravines. Conifer plantations have been implicated in diebacks elsewhere (Struhsaker et al. 1989), and slash pine was heavily planted during the 1950s in the uplands of the southern end of the range of T. taxifolia. Aerial photographs indicate that approximately 36 % of the land within the range of T. taxifolia is forested uplands (Figure 1); most of these forested uplands became pine plantations during the 1950s. The introduced fungus Phytophthora cinnamomi, which causes little leaf disease in southern pines, has been proposed as a

10 Table 1.2. Nine hypotheses proposed to explain the decline of Torreya taxifolia and the data related to assessing these hypotheses.

HYPOTHESIS SOURCE EVIDENCE Introduced Fungus Alfieri et al. 1967 Anecdotal evidence does not support ______hypothesis Pathogen Vectors Schwartz 1990 Very poor evidence

Water Stress USFWS 1986 Climate records support hypothesis by showing a dry periods associated with time of decline. Microclimatic USFWS 1986 USGS water tempera- Warming Toops 198* ture data do not support hypothesis as proposed. Regional Warming new Climate record shows no warming trend, potential cooling of winter temperatures. Does not support hypo- thesis. Hydrologic Changes USFWS 1986 No evidence of the effect of hydro- ______logic changes. Fire Suppression Schwartz 1990 Weak supportive evidence. Air Pollution new No evidence.

Fungal Pathogens as new No evidence. an epiphenomenon______

11 possible introduced pathogen of T. taxifolia (Alfieri et al. 1967, USFWS 1986). This root-rotting fungus, which is thought to have been introduced from Australia (Pratt et al. 1973), is known in northern Florida (Barnard, pers. comm.). Its pathogenicity on T. taxifolia is untested, but has been suggested to be associated with the decline (USFWS 1986). Two pieces of evidence suggest that P. cinnamomi is not associated with T. taxifolia at the present time. First, roots of T. taxifolia that were examined by myself and others in the field appear healthy. Second, Barnard (unpubl. data) examined soil samples from around the base of 17 diseased T. taxifolia. He failed to isolate P. cinnamomi, or any other obvious pathogen, using techniques that had previously been successful in other north Florida soil samples. No other introduced fungus has been postulated as a pathogen. _i1 Changes in Resource use hy Pathogen Vectors Torreya taxifolia suffers stem damage by deer as antler rubs. The conversion of uplands to pine plantations in the 1950s entailed total clearcuts. Earlier harvests of longleaf pine had been restricted to canopy trees, leaving juveniles to regenerate. Deer, which are abundant in the region, prefer soft-barked coniferous trees for antler rubs. The removal of young pines from the uplands may have increased the use of T. taxifolia by deer. In turn, wounds inflicted by deer may have increased the rate of fungal infection and facilitated the spread of disease. While nearly all larger individuals carry scars from past deer rubs,

12 the lack of a pathogen prevents testing of this hypothesis. Cl Fungal Pathogens as Epiphenomenon of Decline There is no guarantee that the symptoms of disease that are now observed on T. taxifolia bear any direct connection to the cause of the decline. Some pathogen, not now found on T. taxifolia, may have swept through the region causing the decline. If such a pathogen is now missing, or at very low densities, we would likely not detect its presence. HYPOTHESES RELATED TO CHANGES IN THE ABIOTIC ENVIRONMENT DI Water Stress Barnes (1985) suggests that drought may have stressed T. taxifolia, making them more susceptible to facultative diseases. Water stress is known to predispose plants to disease (Yarwood 1959, Parker 1965, Cook 1973, Schoeneweiss 1975, 1978, Griffin 1978) as well as weaken a plant's ability to survive fungal infection (Ayres 1978, Burdon 1987). Hepting (1963) and Schoeneweiss (1978) cite numerous examples of secondary pathogens causing stem cankers in birch, maple, and oak during drought years. The foliar fungi isolated from T. taxifolia fit the general description of secondary pathogens (Alfieri et al 1967, 1984, Schwartz 1990). Secondary pathogens have been observed to become primary pathogens under conditions of plant stress (Schoeneweiss 1975, 1978). Weather records from Blountstown show low annual rainfall in 1938 and 1953, years near when disease was recorded in T. taxifolia (Figure 1.2). Although drought may have contributed to

13 Figure 1.2. Annual precipitation recorded from Blountstown Florida, adjacent to the natural range of Torreya taxifolia. The drought year most closely associated with the time of the decline is highlighted.

•JE 90- Blountstown Station

80

70"

Inches 601

40

1953 30- I . •- r •. 192 0 1940 1960 1980 Year the fungal disease, drought seems insufficient as the sole explanation for the decline. The droughts of 1938 and 1953 were not unusual for the period between 1922 and 1982 (Figure 1.2). Because rainfall varies widely in north Florida, adult trees, which live an excess of 100 years, must survive several droughts as severe as those of 1938 and 1953 to reach maturity. EL Microclimatic Warming

Lake Seminole drains into the Apalachicola River at the northern limit of T. taxifolia. Construction of the Woodruff dam, which created Lake Seminole in 1956, approximately coincided with the decline. The dam-related hypothesis is based on changes in

Apalachicola ravine microclimate and flooding patterns (Toops

1981). This hypothesis proposes that water in Lake Seminole warms, raising temperatures in the Apalachicola River. The warmer river, in turn, raises ravine air temperatures in what may already have been a marginal micro-climate for T. taxifolia (Toops 1981).

This hypothesis is subject to verification through direct observation. Water gauge temperature data collected by the United

States Geological Survey (USGS) from above (in the Flint and Chattahoochee Rivers) and below Lake Seminole showed no difference over the 25 year period from which data are available (Figure 1.3). With no change in water temperature, we have no evidence of changes in ravine microclimate that can be attributed to Lake Seminole.

15 Figure 1.3. Water temperature records, by calendar date, for the Flint, Chattahoochee and Apalachicola Rivers. The Flint and Chattahoochee stations are immediately above Lake Seminole, and represent the major inputs to the lake. The Apalachicola River station is immediately below Woodruff Dam. The Flint and Chattahoochee Rivers form the Apalachicola River at their confluence.

* Apalachicola Staton (below Lake Seminose) Data from 1960 - 1985 0 Chaftahoochee Station (above lake Seminole)

A Flint Rivr Station (above Lake Seminoie) 30 I

Degrees 204 (Celsius)

I 10 a

0 * I- Jan Mar May Jul Sep Nov

Month - fgRejional1 Climatic Warming Temperatures in the eastern U.S. were cooler during the 19th century (Wahl 1968). It is possible that the decline was driven by a marginally warmer and drier climate during the 20th century that has made the ravine habitats unsuitable for T. taxifolia. Weather records from the nearby Blountstown station show no trend in mean summer temperature during the monitoring period of 1931 through 1982 (Figure 1.4). Although mean winter temperatures appear to decrease over this period. If climate change did contribute to the decline, then it most likely has been a long continuous slow process that predates the catastrophic die-back of the 1950's. iHydrgloi -change Hydrologic changes within ravines might account for the decline (USFWS 1986). Conversion of longleaf pine uplands to slash pine plantations, during the 1950's, entailed extensive site preparation in the southern third of the range of T. taxifolia. Ravine temperatures may have risen as a result of the highly reflective bare mineral sand exposed from site preparation. Further, the lack of vegetative cover would have reduced evapotranspiration and perhaps lowered humidity in the adjacent ravines (USFWS 1986). Sparse vegetative cover would have altered runoff flow, and soil disturbance overland sheetf low of water. This hypothesis is somewhat unsatisfactory because the extent of upland clearing varied across the range. Large tracts

17 This page is intentionally blank. Figure 1.4. Mean maximum and minimum temperatures for summer (June-August) and winter (December-January) months from 1932 to 1982 as measured in Blountstown, Florida.

100

(1 80

c%,._

CL E -- 60 CO ---- Summer, max. 03 -]--- Summer, min.

40 ---- Winter, max.

-- *- Winter, min.

20 1IW-- Year of uplands adjacent to ravines were not converted to slash pine. We might, therefore, predict that T. taxifolia found in Torreya State Park, or in the outlying population along Lake Ocheesee, would have been less affected by the decline, yet these individuals were susceptible.

HI iU Pollution The establishment of pine plantations in the 1950's was coincidental with the construction of several paper mills in the north Florida region. The pollutants from these paper mills may have reduced the pH of local precipitation. There is no direct information available on how reduced pH may effect disease inception in T. taxifolia. 11 Fire Suppression The suppression of fire, which effectively began in the 1950s, could have had a direct effect on the onset and severity of the T. taxifolia decline. Natural fire frequencies in pineland habitats of north Florida are estimated to occur at 1-3 year intervals (Komarek 1964, 1968, Robbins & Myers 1989). Historical accounts cite the use of near annual fire by both Native Americans and Europeans in Florida (Tebeau 1980, Pyne 1982). This frequent fire regime was halted under fire suppression that began in the 1950s. Although T. taxifolia grows in a relatively nonflammable ravine habitat, it could have been affected by fire suppression in two ways. First, fire suppression allowed upslope portions of ravines to become more heavily wooded. Thus, lower portions of

19 the slopes became more heavily shaded than prior to fire suppression. Increased shading may inhibit recovery through slower growth rates and lower plant vigor (see Chapter 4). Second, fire may have had an indirect effect on ravine habitats by way of smoke settlement. Ravine habitats are lower and cooler than the surrounding uplands. During the evening, smoke from upland fires settles into such low-lying areas. The smoke generated from large fires is extensive, and anecdotal reports cite smoke settling so thick at ground level that one could not drive a car at night. Smoke is used as a preservative of meats and other goods because it inhibits the growth of fungi and bacteria (Parmeter &

Uhrenholt 1975a). Experimental evidence indicates that smoke residue on fungal growth media decreases spore germination and mycelial growth of many fungi (Melching et al. 1974, Parmeter & Uhrenholt 1975a, 1975b, Mihail 1979, Zagory & Parmeter 1984). The deleterious effect of smoke on Fusarium lateritium, the only fungus that has a demonstrable pathogenic effect on T. taxifolia, is particularly pronounced (Zagory & Parmeter 1984). Thus, the frequent and consistent presence of smoke in ravines may have served as a chemical defense and fire suppression may have initiated a fungal outbreak throughout the range (Chapter 5). The smoke hypothesis may fit well if, in fact, foliar pathogens are responsible for the decline. Needle infections do not spread within the vascular tissue of the plant; instead, each needle requires a separate penetration of the stomata by inoculum

20 (Parmeter pers. comm.). Massive infections of needle tissue may be required before trees are adversely affected. Fire suppression in the uplands might have allowed populations of needle pathogens to explode within the ravines. DISCUSSION. Each year new research describes how air pollution, global warming, habitat degradation, and other human-caused changes in the environment can effect biodiversity (e.g., Melillo et al. 1990, Warrington and Whittaker 1990, Woodwell 1990). Discerning exactly how changes in the environment effect species health, however, can be exceedingly complex (Schoenewiess 1975, Bormann 1990). For example, researchers remain divided over the mechanism responsible for the decline of forests in New England and northern Europe (Bormann 1990). The list of fungal isolates of T. taxifolia presented here varies somewhat from previous studies (Alfieri et al 1967, USFWS 1986). Three explanations arise for these differences. First, native trees were used for these experiments, while previous studies used trees growing in the wild as well as trees in lawns, at Maclay Gardens and on the campus of the University of Florida (Alfieri et al 1967, El-Gholl personal communication). These trees outside the natural range are likely to be vectors of other horticultural diseases that may not infect wild trees. Second, Alfieri et al. (1967) list several species only hypothesized to be in association with L taxifolia. Thus, this list of fungal isolates may be more accurate, with one exception. Phyllosticta

21 sp. was previously isolated from ascopores on needles in the wild (Alfieri et al. 1967), and was tested for pathogenicity before with negative results (El-Gholl 1984). Finally, Alfieri et al (1967) did not list fungi that they perceived to be secondary pathogens (El-Gholl, personal communication). The nine hypotheses for the decline of T taxifolia have been summarized (Table 1.2). One hypothesis (microclimatic warming) can be dismissed as a result of the evidence presented here. Three of the hypotheses (introduced fungus, deer vector and hydrologic change) may be related to the conversion to plantation pine during the late 1950s. This conversion focused on the southern third of the range of T. taxifolia, and thus generates a prediction of a specific spatial pattern to the spread of disease. For these hypotheses the disease should have spread from relative epicenters, or had differential effects across the range of the species. In contrast, the four remaining hypotheses (water stress, regional climate change, air pollution and fire suppression) predict a synchronous decline throughout the range. Although anecdotal accounts from residents seem to suggest a synchronous decline, we do not have sufficient historical records to determine if the disease spread from a single locus or was immediately widespread. These nine hypotheses are not mutually exclusive. Further, a single causative agent of the T. taxifolia decline need not be isolated in order to develop a plan for the recovery. However, several of the factors I have described do not lend themselves to

22 management efforts for recovery. Air pollution, regional climatic change, and drought are phenomenon that cannot be locally controlled, reducing the ability to manage for on-site recovery. Past hydrologic changes may affect the future of T. taxifolia in ways that we can neither predict nor control at this time. Finally, it would be difficult to control an introduced pathogen without identification. In contrast, the deer vector and fire suppression hypotheses point to management opportunities for on-site recovery. Whether or not deer act as a vector of disease in T. taxifolia, the antler rubs increase stem mortality. Deer populations can be reduced, alternative antler rub species can be replaced in the uplands, or individual T. taxifolia may be protected using exclosures. Fire management can be restored to the uplands. Both of these factors are currently being addressed through management of local preserves by The Nature Conservancy. This chapter points to several difficulties with respect to species declines. Isolating disease agents, and factors that enhance the virulence of these pathogens, is exceedingly difficult. Yet, the environment faces an ever widening array of environmental changes through air pollution, climate change, continued deforestation, and disruption of natural disturbance processes. The example of T. taxifolia is illustrative because it demonstrates how difficult disease assessment can be. This difficulty in discerning disease agents is enhanced when dealing with an endangered species.

23 CHAPTER TWO The continuing Population Decline of Torreya Taxifolia

Although extinction or near obliteration of populations and species has been of interest to biologists for centuries, attempts to closely monitor and assess such processes are common only during the last few decades. Many of these recent studies focus on endangered species; demographic data on rare plants is increasingly utilized as a tool in conservation efforts aimed at preventing extinction and enhancing existing populations (Davy and Jeffries 1981),o Anecdotal information may suggest the need for closer inspection or hint at possible remedies, but quantified data is needed to accurately assess the health of a population (Travis and Sutter 1986). This may be especially true for long-lived species where the population decline may be slow. In addition, individuals of perennial plants may decline in vigor long before mortality actually occurs. Woody species provide a semi-persistent record of individual decline that can be used to monitor fluctuation at both the individual and population levels.

In the United States there has been documentation of catastrophic decline or extinction of various woody species. One of the earliest examples is of a small tree (Franklinjia alatamaha), endemic to a narrow range in coastal Georgia, that has not been seen in the wild in over 200 years. Over-collecting has been suggested as the reason for its disappearance (Harper

24 and Leeds 1937) but neither the cause nor the pattern of decline was actually documented. Over a much larger geographic scale, eastern forests of North America witnessed the catastrophic decline of the chestnut (Castanea dentata) during the first half of this century. For this once dominant tree species, many components of the near extinction are understood (cf. Braun

1950). In this example, most ecological considerations focused on community level observations related to species replacement (reviewed in Woods and Shanks 1959). At the individual level, only general remarks were usually reported (cf. Nelson 1930). Repeated measures on individuals may be necessary to effectively explore population responses and to document patterns of continued decline or recovery (Travis and Sutter 1986).

STUDY AREA AND METHODS The range of T, taxifolia encompasses several private and public preserves and parks (Figure 2.1). This study focuses primarily on trees within secured habitat, although one site with more than 40 individuals is included from outside preserve boundaries. Each area was systematically searched by multiple workers to insure that all trees within the sample region are included in the census. In 1988, a thorough survey of the ravine systems within the southern portion of the Apalachicola Bluffs and Ravines Preserve (ABRP, Figure 2.1) was conducted. Seventy-nine trees were added to the 25 previously known from this preserve. A repeat census in 1991 relocated 100 of these trees and added six new individuals

25 Figure 2.1. The distribution of TorreyA taxifolia. Highlighted areas show populations included in census or discussed in text.

ABRP Apalachicola Bluffs and Ravines Preserve (TNC) BGSW Big Sweetwater Preserve (TNC) TRSP Torreya State Park FLCR Flat Creek (Private Land) LKSM Lake Seminole (U.S. Corps of Engineers) from a portion of the preserve that had not been previously censused (Table 2.1). Four trees censused in 1988, but not in 1991 are very difficult to find within a remote area of the preserve and were excluded from the permanent census population. In addition, 25 trees from U.S. Army Corps of Engineers land at Lake Seminole (LKSM) were surveyed and tagged in 1988. This was the entire population as reported elsewhere (Savage 1983, T. Patrick,pers. comm.). By 1992, we had added two previously overlooked individuals to this population (Table 2.1). One of these new trees had previously been considered a second stem of a known individual, the other was a newly encountered stem. In mid- summer 1991, 25 trees from a population along Flat Creek (FLCR) were also measured. This census population was increased to 31 trees in 1992. Finally, census information was collected from 27 trees in three separate ravines along the Big Sweetwater Creek (BGSW, Figure 2.1) in 1991. One year of growth information is available for the 16 trees from the BGSW population that were surveyed prior to bud break in 1991. This census population was expanded to 43 trees in 1992. The census population represents all trees encountered along particular stretches of ravine systems within the preserve. The total population on the BGSW preserve has been measured to exceed 125 trees by Nature Conservancy volunteers. A large population from Torreya State Park was excluded from our study. Trees grown from seed were planted in the park during the 1960s and 1970s in an attempt to increase the population of

27 Table 2.1. Number of Torreya taxifolia in the census population during each year of the study along with the number of individuals for which growth history is recorded over this interval. POPULATION A) Census ABRP 1 BGSW FLCR LKSM TOTAL 1988 104 0 0 25 129 19912 106 273 253 24 182 1992 102 43 31 27 203 B) Growth Growth in '89,90 98 0 0 23 121 Growth in '91 106 16 0 24 146 Growth '89-91 894 0 0 23 1115 1 - See text for explanation of site abbreviations. 2 - Increase in numbers between years represent expansions of census areas, not increases in population size within census populations. 3 - Most 1991 census data was collected after new growth and does not include growth information for 1991. 4 - Attrition in numbers is as a result of mortality. 5 - Owing to 1988 census techniques we could not distinguish between stems that grew and stems that did not grow for 20 census trees between 1988 and 1991.

28 T. taxifolia. These plantings were not documented and can no longer be discerned with certainty from natural stock. Tree size was characterized by the number and length of stems, the number of internodes along the main trunk of the longest stem, and the number of branches along the main trunk for the longest stem. Habitat was charactiized-by three measures. Relative elevation above the ravine base was estimated using slope and distance measurements. Canopy density above T. taxifolia individuals were compared to the forest as a whole, using a LICOR light sensor. Four readings were measured above each tree. Ambient light levels were sampled haphazardly while traversing between census trees at approximately 10 m intervals.

Aspect of the slope upon which trees were located was recorded by compass. Plant vigor was assessed through measures of recent growth and presence of disease symptoms. New growth during the census period was recorded as additional internode lengths on primary stems. The length of new growth during the past three episodes of terminal bud extension was also measured. Further, the age of branches was estimated from the number of branch internodes. This measure was compared to the number of stem internodes above the branch point to estimate the frequency of terminal bud extension for each tree under the assumption that lateral branches expand internodes annually (this assumption will be discussed further below). Terminal bud extension frequency allows an age correction factor to be calculated that incorporates the likely frequency of

29 terminal bud extension. The approximate age of the oldest stem on individuals was estimated by counting the number of internodes along the trunk and multiplying by a population-wide estimate of the frequency of terminal bud elongation. To assess disease status we noted whether plants exhibited any of several characteristic disease symptoms. Needle spots are

defined as discrete round patches of dead tissue between 1 and 3 mm in diameter. Needle spots were previously considered the most prominent symptom of foliar disease in the wild (Alfieri et al. 1967). Needle necrosis was defined as dead tissue in a discrete band laterally across needles and toward the needle tip. Stem cankers were defined as any scar along the stem that was not obviously a result of abrasion. These cankers were most frequently swelling from within the woody tissue in larger stems, and cracking and swelling of the bark in smaller stems. Stem scars included all scarring of bark that appeared to be from abrasions such as deer antler r.ubs or treefalls. The presence or absence of disease symptoms, physical abrasions, and scarring was noted for each individual. RESULTS The population of Torreya taxifolia is characterized by a majority of individuals less than 1 m tall (Figure 2.2). The

largest stem for 57% of censused trees was between 25 and 100 cm long, and the mean stem length was 87.8 cm with 6% of individuals having stems more than 2 m long (Figure 2.2). The mean length of the longest stem was not randomly distributed across populations

30 Figure 2.2. The distribution of size classes of the longest stem for individual TorreyA4taxifolia. Data are from trees sampled from throughout the natural range. - 40AJ%

30 U) 0 0 I- 20 0 .0E z 10

In 0 ) a in O n 0 In 0 W0n In 4 In , 0 O4 U, 0 No4 W). 0 N

,.0 N- - ,-t 0 N4 U)4 €0

Length of Longest Stem (cm)

Figure 2.3. The number of stems per individual TorreaA taxifolia. Data are from trees sampled from throughout the natural range.

80

U) 0 60 0 I.- 40

S z 20

1 2 3 4 5 6 7 8 9 10+ Number of Stems (Table 2.2) and was larger toward the north portion of the range. This trend appears to be a result of more very small individuals toward the southern range limit (Table 2.2). This pattern of increasing size with latitude was repeated within ravine systems of the southernmost population (ABRP). Most individuals are multi-stemmed, although the most frequent number of stems per individual was one (Figure 2.3). Length of longest stem was positively correlated with stem number, although the relationship was weak (r2=.19, p<.001). Thus, it is not surprising that individuals in the more northerly populations have more stems (Table 2.2).

Torreya taxifolia is most commonly found at low elevations with respect to ravine slopes (less than 6 meters above the ravine base), although they may range well upslope (Figure 2.4). The ravine habitats of T. taxifolia span any where from 12 to 45m in elevation above constant running streams at the base of each ravine (Schwartz 1990). The substrate of lower slopes is characterized by higher soil moisture and higher organic content than that of upper slopes (Schwartz 1990, unpublished data).

Torreya taxifolia is also generally found under a relatively closed canopy duringthe summer (Figure 2.5), although this pattern is descriptive of the habitat in general and does not imply habitat selectivity in T taxifolia. Measures of photosynthetically active radiation (PAR) averaged 215 Mmol/m2 /s (~s.d.=284, n=257) adjacent to T_. taxifolia, compared to 254 Mmol/m2/s (s.d. 308, n=293) for random points in the forest

32 0 (U U) 0 4.) .1 (U H p :I U' LALANO 0 p4 4w1 *1I

* 1 1%0%O O O*%** S 00 to I r% co t-0 IC

-44 I '0 x I (UI I 00 14 0 41I 10011T 0 I CO I CO C- 01 0C-4* 0 1I 0 0 coN tkHI(q * 0 0 I 4.) I * en U>I

IIV4%r~4P cl 0 I 0 IA I r .4rZ (1)~ I 4.)4.)44P 4JI 1 41 00 U) *% I

(U HI 00 41 AI 0 WI *I 4F40 4404. ~~~~I-r.4J

44 V4A U00I 431 0' LO I U'$4( 0. IC4H0 U)M 4I > I 9H04 4 0r4~ I lQ0

4J .~0

04) r- c 4z ~4J

E-~4 Figure 2.4. The distribution of Torreya taxifolia across relative elevation above ravine base. Data are from trees sampled throughout the natural range. Ravines vary in depth from 12 to 45m in relative elevation.

301

0 20

1 10 z

0 1 2 3 4 5 6 7 8 9 1011 1213+ Relative Elevation (m)

Figure 2.5. The distribution of Torreya taxifolia by canopy cover class. Data are from trees sampled throughout the natural range. Canopy openness was measured as the percent of 100 random points that did not intercept canopy vegetation in overhead 28mm wide- angle photographs taken above individual trees.

80-

40- in wuwier W

0 20.0

Canopy Openness

Figure 2.6. The distribution of Torreva taxifolia with respect to aspect. Data are from trees found along slopes sampled throughout the natural range.

401

S 30 0- "S 20

z| 10

0 Aspect during leaf flush in March. These values averaged 25 % of full sun levels measured during this period (x=1043 gmol/m 2 /s, s.d.=547, n=42). Incident levels of light is correlated with slope position; lower slopes tend to receive lower incident light levels because of increased shading from uplsope vegetation as well as an increased angle to the horizon (Schwartz 1990). Finally, T. taxifolia is distributed widely with respect to aspect (Figure 2.6). The distribution of the apparent age of primary stems, as assessed by the number of internodes, ranges between 1 and 21 years, with a mean of 8.0 (Figure 2.7). True stem age is related to apparent age by the frequency with which stems expand an apical bud and grow new internodes along the main stem. We have observed that stems do not expand apical buds each year. The relationship between apparent lateral branch age (lateral internodes) and the number of terminal internodes above that branch point provide a correction factor for age. This age correction factor (Figure 2.8) suggests that terminal buds are expanded approximately every other year under the assumption that side branches expand leader buds each year. This correction factor suggests that the mean age of stems is 12.0 (s.d=6.3), with maximum stem age at 31 years. However, over the three years of growth encompassed in this study (1989-1991) only 47% of surviving trees expanded any terminal buds (see below), while approximately 20% of trees expanded a new terminal internode each year. This low rate of terminal bud expansion during the census

35 iqgure 2.7. Apparent age distribution of the oldest stems of individual Torreva taxifolia found in the Apalachicola Bluffs and Ravines Preserve. Apparent age is estimated from counting the number of terminal internodes along the main stem.

30

20

0 I- 10 z

Apparent Age

Figure 2.8. The distribution of the number of termanal internodes above individual branch points plotted against the number of internodes along the branch eminating from that branch point. This scatterplot estimates the frequency of terminal bud elongation under the assumption that lateral branch buds elongate every year. Thus, the relationship described by the these points provides a correction factor for determination of the approximate minimum age of individual stems.

12 * 10 I ** U U 8- e W II 6- I * •Ir • .lU I 4. * 0 0 r 2 = 0.89 2 * Ya0.47 + 1.4 s5*X U

u U - I - I 0 2 4 6 8 10 NUXMs or TRanzL IrTnoo1s interval indicates that our corrected stem age estimates are also low. Thus while we cannot age specific stems, the census population appears to consist of a mixture of trees whose primary stem dates from the time of the decline along with trees whose primary stem is a sprout from an earlier stem that has died since the onset of the decline.

Several patterns of disease incidence are noteworthy. First, needle spots are found on nearly all trees, but rarely found to exceed 5% coverage of needles (Table 2.3). Second, needle necrosis, which has not previously been reported in the literature, is widespread at low levels throughout the population (Table 2.3). Finally, stem cankers appear to increase in more northerly populations. However, this pattern may simply reflect the fact that cankers often do not manifest themselves in smaller stems. Canker occurrence is associated with longer total stem length (t-test, T=4.43, p<.001). Needle disease incidence also varied considerably between populations (Table 2.4), but small population sizes at several sites make it difficult to accurately test the significance of this variability. The data from this study record growth and mortality for 111 individuals in two populations (ABRP and LKSM) during three growing seasons (Table 2.1). During this period 11 individuals (9.9%.) from these two populations died. All mortality was among individuals from ABRP and ranging from 17 cm to 101 cm tall (x=46.9 cm, s.d.= 23.6, n~ll) with two or fewer stems. Of the 117 plants that survived the 1988-1991 census interval, 18 (15.3%)

37 Table 2.3. The percent incidence of disease symptoms and damage in Torreva taxifolia. Stem cankers are lesions in the outer cambium visible from the surface. Stem scars are any scar tissue along stems caused either by deer or any other abrasion. Needle spots are brown circular spots that are distinct and isolated from other damage. Needle necrosis are dead ends of needles that constitute more than one-third the needle area. SYMPTOM POPULATION Population ABRP BGSW FLCR LKSX mean

Stem cankers 26.1 46.5 64.5 52.0 40.3 Stem scars 27.2 58.1 6.4 4.0 27.7 Needle Damage 79.1 90.3 100 85.5 Spots >0% 81.9 52.2 36.3 >1% 35.1 32.6 25.8 7.3 >5% 7.4 14.0 0 4.3

Necrosis >0% 57.4 44.2 37.5 24.0 45.4 >1% 29.8 11.6 0 12.0 17.1 >5% 9.6 4.7 0 0 5.7 N 94 43 31 25 193

Table 2.4. Number of trees, by size class, differing in growth fate from 1989-1991. Values in parentheses indicate percentage of size class total (row total). Size class Grew Did not Stem Died Total grow died

1989, 1990 0 - 50 cm 10 (34%) 10 (34%) 7 (24%) 2 (7-%) 29 51 - 100 cm 16 (38%) 17 (40%) 7 (17%) 2 (5%) 42 101- 150 cm 6 (38%) 7 (44%) 3 (19%) 0 (0%) 16 151- 200 cm 4 (36%) 7 (64%) 0 (0 %) 0 (0%) 11 > 200 cm 1 (14%) 5 (71%) 1 (14%) 0 (0%) 7 N 37 (35%) 46 (44%) 18 (17%) 4 (4%) 105 1991 0 - 50 cm 8 (14%) 29 (50%) 17 (29%) 4 (7%) 58 51 -100 cm 13 (21%) 38 (62%) 8 (13%) 2 (3%) 61 101- 150 cm 3 (15%). 16 (80%) 0 (8%) 1 (5%) 20 151- 200 cm 7 (37%) 12 (63%) 0 (0%) 0 (0%) 19 >200 cm 4 (44%) 5 (56%) 0 (0%) 0 (0%) 9

N 35 (21%) 100 (60%) 25 (15%) '7 (4%) 167

38 lost their primary stem and decreased in size (Table 2.4). Likewise, of the 182 census trees that survived between 1991 and 1992, 25 (13.7%) lost their primary stem and decreased in size. When trees lost their primary stem they decreased an average of 45.2 cm (s.d.= 52.6). In total, 32% of those trees followed over the entire four-year census interval lost a primary stem during this period. Differences in stem lengths between 1988 and 1991, along with number of internodes on the main trunk, were used to determine which trees expanded terminal buds during the census interval. Because the 1991 census was completed prior to bud break in 1991, the 1988-1991 census interval comprised only two years of growth. For several plants it was unclear whether a terminal bud had expanded. It-was difficult to accurately measure the tops of the taller trees, and many of these questionable plants are among the larger stems. Between 1991 and 1992 growth fate was more accurately determined through a comparison of recent terminal internode lengths to determine if a new internode was added. The most frequent fate of trees over both census intervals was no new growth (Table 2.4). Approximately 35% of trees grew between 1988 and 1990, while 21% of individuals grew in 1991. Over the census interval not one single tree expanded terminal buds in all three years, fewer than 20% grew in two of three years, and only 47% expanded a terminal internode at some time during the period. The only pattern between size and growth

history is that the majority of stem and individual mortality was

39 among small individuals (Table 2.4). Average length of new growth among those trees that expanded a terminal bud was 9.60 cm (s.d.= 5.2). In addition, there was a weak, but significant, positive linear relationship between the length of the primary stem and the amount of new growth during both census intervals (r2 < 0.30, p S .02). Among the 13 trees that lost their primary stem between 1988 and 1991 and expanded terminal buds on secondary stems average two-year growth was 13.6 cm (s.d.= 8.7 cm) on the new, shorter primary stems. Combining growth and mortality results for those trees followed for four years we find that the mean length of the primary stem decreased by 2.5 cm over the census interval. This decrease in mean size is accounted for by height loss associated with stem death, coupled with sparse gains in height through growth and a slight positive effect in mean size through differential mortality among small individuals. Trees that grew in 1991 had fewer stems (x=--2.4, s.d.=l.l, n=36) than those that did not grow (x=2.7, s.d.=2.0, n=99) or whose primary stem died (x=3.0, s.d.=2.3, n=25). This pattern was repeated from the 1989, 1990 growth data. Trees that grew in 1991 also more frequently exhibited stem cankers, although this pattern is probably a result of the fact that cankers are more frequently found on larger stems, and that larger stems more frequently grew in 1991 (Table 2.4). Trees whose stems died during 1991 routinely exhibited a higher incidence of moderate levels (>1%, >5%) of needle spots and needle necrosis (Table 2.5). Finally, trees that grew and were free from needle necrosis

40 grew more (x=9.8 cm, s.d.=6.0, n=17) than those that exhibited needle necrosis (x=6.5 cm, s.d.=3.2, n=19)(T-test, T=2.02, df=23.6,p=.05), although this pattern did not hold for needle spots.

G-tests of growth category (Grew, Did not, Stem Died, and Died) by habitat categories (relative elevation, canopy cover, and aspect) revealed no significant patterns. Similarly, the occurrence of stem cankers, needle spots and needle necrosis were unrelated to habitat variables. Finally, growth fate of trees prior to 1991 was not related to growth fate during 1991 (G-test of independence, G=1.78, df=7, p>.5).

DISCUSSION This work suggests a refinement in the estimate of the number of extant T. taxifolia. This survey, along with additional census information collected by Nature Conservancy volunteers on their preserves account for 240 trees. An unpublished report from

Torreya State Park includes over 200 trees (Bryan, personal communication). Several smaller populations not included in this survey account for nearly 100 additional trees. Thus, we can directly account for over 500 trees. This number exceeds one estimate of the remaining population (Stalter and Dial 1984).

Most known populations of T. taxifolia are within 2 km of the Apalachicola River. This study, the Torreya State Park survey, and Baker and Leonard (1982) censused most ravine habitats within

2 km of the river in the southern half of the range of T. taxifolia. The northern populations in this study (LKSM, FLCR)

41 Table 2.5. The percentage of trees, classified by growth fate in 1991 exhibiting various symptoms disease. Needle spots and needle necrosis are categorized by the portion of the total green foliage affected by the symptom.

Growth fate Disease symptom 1991 Stem Stem Needle Needle canker scar spots necrosis >0% >1% >5% >0% >1% >5%

Grew 44.4 25.0 94.4 30.6 5.6 52.8 19.4 0.0 Did not grow 35.4 27.3 76.8 33.3 3.0 39.4 13.1 3.0 Stem Died 20.0 24.0 92.0 60.0 24.0 68.0 48.0 28.0 Population 35.0 26.3 83.1 36.9 6.9 46.9 20.0 6.3

42 are included because they contain the only large populations of trees known from this portion of the range. Based on our assessment of the thoroughness of recent surveys, along with anecdotal accounts of populations throughout the range (Baker pers. comm., Gholson pers. comm.) we estimate that the extant population is probably between 800 and 1500 individuals, and almost certainly does not exceed 2000 individuals. The pattern of larger mean size of individuals toward the northern range limit may be a result of differences in habitat.

The Cody escarpment, delineating the Pleistocene sandy soils in the south from the clayey Miocene soils in the north, divides the range in the Big Sweetwater Preserve. Soil differences across this escarpment may result in differential abilities of T. 1axifolia to survive and grow. Alternatively, this nonrandom assortment of stem sizes may be a result of differences in the geographical distribution of foliar disease, as suggested by Table 5. One aspect of the demographic pattern of the T taxifolia population that is not depicted well by these data is that the trees are less abundant within the appropriate habitat in the northern portion of the range. This is reflected somewhat by sample sizes within populations.

The trunks of dead T. taxifolia remain well preserved on the forest floor. The distribution of these trunks, as well as anecdotal accounts, suggests a formerly more widespread distribution with respect to slope position than was observed in this study. Further, previous accounts of the species suggest an

43 historical distribution that included more trees further up ravines away from the Apalachicola River (Figure 2.1). Few of

these trees remain. Thus, the decline seems to have been accompanied by a narrowing of the environmental conditions across which T. taxifolia may be found. Also of note with respect to dead trunks on the forest floor are our personal observations that these trunks were never associated with live T. taxifolia. Thus, the current population does not appear to consist of survivors from the decline but probably grew from seeds germinated during or after the onset of the decline. This observation is contrary to other reports implying that the current population consists of stump sprouts from formerly large individuals (Toops 1981, Savage 1983, USFWS 1986). It is not known how long seeds may lie dormant in the soil, but a large nut is characteristic of seeds that do not remain dormant for a long period of time.

Incidence of disease symptoms also varies geographically across the range. High levels of foliar disease symptoms (needle spots and necrosis) appear to be more frequent on trees in the southern half of the range, while cankers are more common on trees in the northern portion of the range. However, interpopulation patterns of disease incidence is confounded by other measures of plant vigor. For instance, stem cankcers are more readily discernible on larger stems. Therefore, we expect a higher incidence in populations of larger trees. However, these results provide evidence that stem cankers are• not associated

44 with decreased plant vigor, as measured through frequency and amount of new growth on cankered versus uncankered stems, while symptoms of foliar pathogens are associated with higher rates of stem death and less growth in the case of needle necrosis. However, needle spotting does not currently appear as chronic as suggested previously (Alfieri et al 1967, 1984, USFWS 1986).

The growth data demonstrate that the remaining population is continuing to decline with 10% mortality, 32% stem mortality and mean size of trees decreasing by 2.5 cm during the four year census interval. The mean age of the oldest stems of individuals is approximately half of what we would expect if the population represented new, healthy recruitment following the decline in the 1950s. Thus, most individuals have lost their primary stem at least once since, the decline 30 years ago. Further, the age estimates for the census population approximate a normal distribution, rather than being clustered at any single age, suggesting that continued mortality among primary stems is inhibiting maturation of the population. However, some of the larger trees (>2 m) appear to represent first-generation recovery from the time of the decline. These census data allow an assessment of the pattern of decline in T. taxifolia. Individuals that died during this census period were smaller than average. The average length of a primary stem that died on a surviving individual was slightly longer than the average length of all primary stems, leaving a new primary stem that was shorter than the mean for the population. Although

45 few individual cases of stem mortality can be attributed to any

single factor such as stem girdling, we can speculate that the the mortality figures results highlight two modes of stem

mortality. First, small, suppressed, and presumably stressed, plants eventually die when reserves are depleted and they cannot maintain sufficient physiological activity. Second, older and longer stems appear to eventually succumb to progressive disease or fall prey to girdling by abrasions such as deer antler rubs, leaving smaller stems behind. The fact that trees that grew had fewer stems than those that did not implies that resprouting behavior may also be a sign. of stress.

The true measure of whether T. taxifolia is exhibiting greater than expected levels of mortality lies in a comparison with other woody species with which it is associated. Stem mortality was just over 10% for all trees greater than 2 cm in diameter at breast height during a three-year census of trees in a mapped plot in the same region (Schwartz 1990). Although these data are not directly comparable, individual mortality (10%) and stem mortality (32%) among T. taxifolia appears higher than expected. This high mortality rate combined with the lack of net growth insures that the population will continue to shrink both in numbers and mean size. There is hope of a recovery, however, from a small subset of trees (<10) that seem to exhibit few Symptoms of disease and

appear to be approaching maturation. At present, there are four sexually mature males and no sexually mature females known in the

46 wild. Two trees are large enough that we would predict them to be sexually mature if they were male,, indicating that these may be females. Without direct intervention to facilitate growth and survival, T. taxifolia appears destined to continue its slow decline toward extinction in the wild. The problem remains as to what caused the decline and whether the vector still persists in the current population, if any management intervention can increase growth and vigor among the remaining individuals, and whether those few individuals that appear to be doing well are so because they are resistant or because they are lucky.

47 CHAPTER THREE Genetic Variability in Torreya taxifolia

As a result of potential imminent extinction, the stinking cedar was chosen a target species for ex situ conservation by the Center for Plant Conservation and the Arnold Arboretum. Isozyme analysis was used to characterize the genetic structure of T. taxifolia in order to develop a plan for collecting and managing material for a permanent collection to be housed in botanical gardens. Studies of the genetic variation associated with rare species are important for developing our understanding of the genetic structure of rare species (e.g., Falk and Holsinger

1990), as well as developing general expectations of levels of genetic variation in species that differ life history characteristics (e.g., Hamrick and Godt 1989). Assessing genetic variability in T. taxifolia will assist in conservation efforts to save this particular species in several ways. First, measuring the amount and spatial distribution of inter- versus intra- population variability will help determine a management strategy for ex situ conservation. Second, this genetic characterization of T. taxifolia can potentially identify isozyme markers associated with disease susceptibility or resistance. Finally, this study will help to determine whether the on-going decline has caused a detectable genetic bottleneck.

48 METHODS

Needle tissue was gathered from 189 trees in 17 populations distributed among six major drainage regions across the range of T. taxifolia (Figure 3.1). In each population we sampled all individuals encountered up to a maximum of 31 individuals. Table

1 provides sample sizes from each population. Most (150) trees sampled for genetic analysis also had cuttings removed for

captive propagation. All sampled trees were individually tagged for future reference. The measure of genetic variability used for this study was starch gel electrophoresis following the techniques of Conkle et al. (1982). Starch gel electrophoresis allows detection of allelic variation from needle tissue (Conkle et al 1982). Foliar samples were processed and analyzed in the USFS Tree Genetics

Laboratory in Berkeley, California in collaboration with Dr. C. Millar. Twenty loci among 16 enzyme systems were identified (Table 3.1).

For each drainage region, I present the sample size, mean number of alleles per locus, Nei's (1978) unbiased estimate of expected heterozygosity (He), and the ratio of observed heterozygosity (Ho) to He . Between drainage, and between population, genetic distances are reported using Nei's unbiased genetic distance (Nei 1978). The total genetic diversity (Hi) at each locus was partitioned into four hierarchical components (Nei

1977): variation within populations (Hp), variation among populations within drainages (Dxd), variation among drainages

49 Figure 3.1. The range of Torreya taxifolia depicting drainage regions in which trees were sampled for genetic analysis. These sampling units are: A) Lake Ocheesee, B) U.S. Corps of Engineers, Woodruff Dam, C) Flat Creek, D) Torreya State Park, E) Big Sweetwater Creek, F) Apalachicola Bluffs and Ravines Preserve. within regions (Ddr), and variation among regions (Dr). Regions

described in this hierarchical anlysis are: 1) drainages north of the Cody escarpment, where relatively clayey soils dominate

(Figure 3.1, Lake Seminole, Flat Creek, Torreya State Park and Big Sweetwater drainages); 2) drainages south of the Cody

escarpment, with relatively sandy soils (Apalachicola Bluffs

Preserve); and 3) trees from the outlying population not within any ravine drainage (Lake Ocheesee).

The genetic relationship between populations was further examined using a cluster analysis of Nei's unbiased genetic

distance. Cluster analysis used the unweighted pair-group method

with arithmetic averaging (UPGMA, Sneath and Sokal 1973) on the 17 populations sampled. All calculations were completed using BIOSYS-1 ver. 1.6 (Swofford and Selander 1989). RESULTS

Of the 20 loci sampled, only seven exhibited allelic variation in the trees sampled, three of which contained

variation at a single tree or in a single population (Table 3.1). Further, each variable locus contained just two alleles (Table 3.1). This low level of variability resulted in few polymorphic loci and low heterozygosity (Table 3.2). The mean genetic distance between populations within drainages (x=0.012, range=0.000 to 0.044, n=29) is equal to the mean genetic distance between drainages (x=0.012, range= 0.001 to 0.028, n=15) and slightly, but not significantly, lower than the mean of genetic distance between all populations (x=0.016, range=0.000 to 0.062,

51 00 r%.0 0 00 V-4I0 4) %00 m~oo I 00 00 e0 0 00 .0 C~00 F-I .00 .0 00 00 Ho0 Lin LO e.)'3CV) C~00 Ho z~00 00 LALA~w .00 00 .0 %0)00 %0 1 V-4 ON 0 .0 0 VH4o 00 rqc C~00 '0 0 00 00 0 0 000 00 00 H00 c00 (~00 hIt * 0 00 00 00 00 c.00 C.00 '4J '00 LO N-4 r-40 Hno 00 0. H00 00C.' 000 00 00 00 1 CC4e 0 00 ~J00 H. 0 00 q~00 (%00I1 00 00 H.0 0 0 00 001 '000 00 $4- .0 -40 00 0 00 -I . 00 Ho %4~ 00 00 t~00 0 0 001 00 H 00 00 4000L LA 0 %0I 00a 04 00 000. coH LOJC IV H00 00 001 I00 0 0 0 0 co0 001 04 H~ Ln LA z 00 HON 00 000 0 LnmI N\ co H H.0 0 0 (0 co ICO LAOLH 0 0 004 04~ 4) 00Lq ~000 00 v 00 0 00 0 0 00 T-4c~ 0%C% H 00 0 0 '0 0 0 0 LOLO 0 0 0 r1-4 0 0 u-00 r4-

HO 0 r-4 0.- 0# E4)

to V-4 4IJ I-I .9.4 ei4Z to 0 )0E--4eSJ

I I I I I I I I I I I O CO I II 0O co 4 I I4J>94 1I %0 CM cor-I4 0iw 1-4m0 COw I *4 1 f4 Co * * * I lU)0 I * * *I IOE lwI I IO I I 1 t 1 I IN I I 10 I I I M I I 1 0 I - ~ I 4-) owI%* %-O 0 -- -- % * 1 0 4 1I N oLA mHM> 4I I 1 0000n00n0000 I 0 1 0oo00 I ooo001o 0 0 I *4 I I I I I I 0 I I I 0 I U I I 0 I *d I I IU*4,C I I 0 I 04 I I 0O II I 0O II

I z* 4) 1 00I I 4J IOH I I I I o 4 I I Z 4) I I *0U)I I 0 I 0"s6 I I t9oI u-u4I * * * * * * * * I

'4

I OI WH H M M M MI

*Q I O 9rI I 0 .O- I r0I w- w I ' w 4)0 9 I OU) I :I 4-U II I I 4.01 I 4 ) 4 4)I I I I I 4.). 4) 4 4) I 0 0 10I ** *00 * * *** *** 0 *m*N *** I I 10 -I 4) . II N IH IU 00 0 I P 9 10 .)I I P 10* 0 ~I 04I I I.0 0 0 4.) 0me I I 4 4H 10 **M 0 1**1 U)** 0 * I *-4 I 10a U) UI. eQ 0E I > U) U I E4 10 0 4. 0 0 I I0 I0b O 0 .4 0- I I 1, 0 *M 0 4 I0 I LA r4 'QMe LA''I I 1-I > , n=136) (Table 3.3). The mean genetic distance between the outlying Lake Ocheesee population and all other populations is relatively high (x=0.018, range= 0.000 to 0.050) (Table 3.3). However, this estimate of genetic distance is weak because sample size is small. Only five trees remain in this outlying population. Using a hierarchical analysis to partition the total genetic variation sampled (H.) into variation found at the population, drainage, and regional levels demonstrates that most (79%) genetic diversity is at the population level (Hp). Of the remaining variation, relatively more was partitioned among populations within drainages (Dpd, 10%) than either between drainages (Ddr, 6.5%) or between regions (Dr, 4.5%) (Table 3.4). Supporting the prevoius results, the cluster analysis of Nei's genetic distance did not discern evidence of structured dif- ferentiation among populations, drainages, or regions (Figure 3.2). DISCUSSION Plants with similar life-history characteristics often have characteristic levels and distributions of genetic variability

(Hamrick and Godt 1989). Our observation of low genetic diversity in T. taxifolia, along with this diversity being distributed within, rather than between, populations fits the general pattern of the distribution of genetic diversity in obligately outcrossing, wind-pollinated coniferous trees (Hamrick and Godt

1989). However, genetic variation among conifers with extremely

55 4) 1 P HI 0

IID 0 000'K00 10 0 4) 1 f-4 00 00> I II U4J0 1 1 IN (4 0%0 001w IrzvoI 001000104'00N01 004to I v-4 IM*1 * 00 0 * 0 1

04 INIV'K 0H0e I HO Irzr'I 'K00001

31 O0N Ia ION I 4-)I '0000 00 1

000 10 0I 0 I

4J I I * V-' vM0 %DN0'a1 4.10 I,-4I000000001

w cI 1-000000000.1 11II -) 0v I I 0OW0000WON I 4J I0 N Vc40 N00 V-40 0 N I OH 4) 1O H 0 000 If-1000 'IG0'L00000 0 f4I1 4.)$.M I 1 .0.000' 0 H41 0 00 0 I

fl'O i za LI1 0 00 00000000 1

(0 1 I I ( 000'K 'D~-~0000 0 0 I

*0 I I'ONu- O OC tv I u0I 000 I

~I I K000~ 00 OakI a ') ' 00 - 0 00 0 H Table 3.4. Estimates of the partitioning of the total gene diversity (Ht) for seven polymorphic loci. See text for explanation of levels of genetic variation.

Locus H D Ddr Dr PGI-2 .488 .3 8 .*09 -.011 .061 AAT .496 .387 .043 .064 .002 SIX .326 .280 .010 .032 .004 UGP .046 .043 .005 -.001 -.001 IDH .019 .016 .000 .005 -.002 MDH-2 .013 .013 .003 -.002 .000 PGI-1 .002 .002 .000 .000 .000

Mean .199 .157 .020 .013 .009 Percent 78.9 10.1 6.5 4.5

57 Figure 3.2. Cluster Analysis of 17 Torreya taxifolia populations using Nei's (1978) unbiased genetic distance.

.05 .04 .03 .02 .01 .00 Drainage Population Region Torreya St Park A D

:Torreya St Park B D

'Big Sweetwater A B

--- Beaverdam Creek F --- - Ocheesee I I A

SIAspalaga Creek D

--- Big Sweetwater D B

I I III 'No Name Creek A F Lake Seminole B

- No Name Creek C F ------'No Name II Creek B F

- Kelly Creek F

- Big Sweetwater E E

-- Big Sweetwater B E

--- Big Sweetwater C E

Flat Creek B C I Flat Creek A C .05 .04 .03 .02 .01 .00

Farris (1972) "F" = 127.4 Percent Standard Deviation = 98.7 Cophenetic correlation = -0.521

58 limited distributions, such as torrey pine and, can be variable. For instance, torrey pine has very low variability (Ledig and Conkle 1983), while monterey pine has high levels of isozyme variability (Millar et al. 1988, Moran et al. 1988). Both species have severly limited distributions and small populations. This characterization of the genetic diversity in T taxifolia is consistent with a pattern for a species of tree that has been subjected to population bottlenecks (Critchfield 1984). A genetic bottleneck hypothesis is supported by the fact that T. californica, the only other North American congener, has relatively high levels of genetic variability (Schwartz and Millar, unpublished data). The observation that there is little differentiation between populations suggests that this hypothesized genetic bottleneck occurred prior to the current decline. If the depauperate nature of T. taxifolia were a result of the current decline, we would expect to see few common alleles and many rare alleles scattered among the individuals sampled, which is not the case. The observed pattern of genetic variation, under a hypothesis of a current genetic bottleneck, is only likely if the inferred isozyme alleles lost as a result of this decline were tightly linked to characters that were strongly selected against during the current decline, a scenario which seems improbable. The hypothesis of a past genetic bottleneck is supported by two additional pieces of evidence. First, Taxus floridana, a small coniferous tree restricted to nearly an identical range as

59 T. taxifolia, also has uncharacteristically low levels of genetic variability (Schwartz and Millar, unpublished data). Second, isozyme analysis of a few individuals that have been growing outside the range of T. taxifolia since before the decline do not show higher levels of genetic diversity than trees in the wild (Schwartz, unpublished data). Thus, T. taxifolia probably experienced at least one other population bottleneck prior to the current population decline that resulted in a reduction of genetic variability in the species. This is not to say, however, that the current decline has not resulted in significant loss of genetic variation. Given the magnitude of the current decline, it quite likely has suffered additional loss in genetic variation. Finally, with the low levels of genetic variation observed in T. taxifolia, it is improbable that a supposedly neutral marker, such as an isozyme allele, could be identified that would identify particularly resistant, or susceptible trees. Indeed, none of the three rare alleles .sampled in this study were found in either of the two trees sampled that appear to be healthy and maturing in the wild. Given that we have not been able to identify which, if any, of the disease symptoms currently exhibited in the wild are associated with the primary pathogens of the decline (Schwartz 1990), testing for the co-occurrence of alleles and disease symptoms is premature. Management Recommendations The adopted strategy for ex situ conservation of T. taxifolia has been to place one cutting from each of the 150

60 trees sampled at each of four botanical gardens (R. Nicholson. Personal Communication). These cuttings will then be reared to maturity. This strategy was adopted to minimize the risk of loss of any individual genotype. Given the low inter-population variability of T. taxifolia, it appears unlikely that the crossing of individuals from different populations will create outbreeding depression problems, or disrupt locally adapted gene complexes. Thus, this management course should be maintained. Further, a few mature trees from four known locations outside the natural range are currently producing seed. Inclusion of seed from these botanic garden and yard trees could accelerate the process of developing a permanent ex situ collection, as well as assist in developing a population for eventual re-stocking of native habitat. From the results presented here, it appears unlikely that inclusion of this genetic stock of unknown origin would significantly disrupt the ex situ conservation population. Cultivation from available seed is recommended following genetic screening of these reproductively active trees.

61 CHAPTER 4

The Light Relations of Torreya taxifolia with Special Emphasis on the Relationship to Growth and Disease

Light is often a limiting resource for plant growth (e.g., Tilman 1988). Low light conditions can create plant stress, potentially enhancing the ability of pathogens to successfully attack plants (Manion 1981). One hypothesis that may explain the lack of recovery in T. taxifolia is that growth may be limited by chronically low light levels (see Chapter 1). This hypothesis

relies on two conditions. First, that there was a decrease in light level penetrating to lower ravine slopes associated with fire suppression in the surrounding uplands and upper ravine slopes (see Chapter 1). Second, the distribution of the species has been compressed to lower slope (and hence lower light) positions; a condition witnessed by the locations of dead stems

on the forest floor as well as current mortality patterns (see Chapter 2). Thus, low light may be associated with decreased

survivorship, reduced growth, and increased disease symptoms. These phenomena relate to the hypothesis that increased shading through increased densities of hardwoods on upper slopes, as a result of fire suppression, could have contributed to decreased vigor and the decline of T. taxifolia (see Chapter 1).

62 In addition, the foliar pathogens causing needle spots and needle necrosis (see Chapter 2) may inhibit T.*taxifolia's ability to thrive in the low light conditions in which it currently grows. More specifically, foliar pathogens may inhibit maximum photosynthetic rates in T. taxifolia. The relationship between foliar pathogens and plant vigor is particularly important in T. taxifolia because a causative agent of the catastrophic decline has not been ascertained. One hypothesis is that foliar pathogens were responsible for this decline, and are responsible for the current lack of recovery in the species (see Chapter 1). We can begin to test this hypothesis by examining the effect of foliar pathogens on rates of photosynthetic activity. Alternatively, several phenomena associated with plant health can be addressed by studying the light relations of T. taxifolia. Hypothesized environmental stress, such as increased ravine temperatures, decreased humidity, and drought, may have increased plant stress in T. taxifolia, and enhanced disease. An obvious symptom of this purported plant stress would be a decreased ability to photosynthesize under stressful conditions.

METHODS We use two general procedures to'assess the light relations of T. taxifolia. First, we assess the effect of light levels on tree growth ththrough field surveys and experimental manipulation of the light environment of trees. Second, a portable gas exchange system (LICOR-6200) was used to measure photosynthesis in T. taxifolia under varying environmental conditions. For these

63 assessments of physiological response to environment stress we assume that photosynthetic rate is a measure of plant vigor and that decreased photosynthetic rates indicate plant stress. Light and Growth To assess the relationship between light level and growth a correlation between recent growth and canopy cover was used. This data was coupled with an experimental test of whole plant growth in canopy openings versus growth under closed canopies. Twelve trees were selected for use in an analysis of recent growth and light levels. All trees were between 95 cm and 328 cm in height. Recent growth was measured through three variables: 1) mean length of the three most recent terminal internodes along the main stem, 2) maximum terminal internode length of these three most recent terminal extensions, and 3) the proportion of lateral branch tips extending new growth in the current year. These growth values were compared to an estimate of canopy closure for each tree. Canopy closure was measured using the frequency of point light intercepts assessed through overhead photography. Four photos were used to characterize the canopy over each individual tree: one directly overhead, and three at 450 angles facing due south, east, and west. The northern aspect was excluded as it is less important in determining the amount of daily direct sunlight a plant receives. Slides of the canopy were projected onto a screen containing 100 randomly located points. The same 100 points were used for all slides. For each slide, the proportion of points where sky was obstructed by vegetation was

64 recorded. A mean of the four slides was used as a measure of

percent canopy closure. Canopy closure was correlated to each of the growth measures through linear regression.

The experimental test of the effect of light levels on growth was begun in the spring of 1991. Twelve trees were chosen to have canopy gaps opened above them. All trees were located on the Apalachicola Bluffs and Ravines Preserve (Figure 2.1). Terminal bud expansion, and the length of the most recent needle bearing branch internodes along the main stem were measured as response variables. These response measures were compared to measures of plant growth produced prior to the canopy opening. To assess the relationship the presence of disease symptoms and light level we used a contingency table test (G-test, Sokal and Rohlf 1981) between visual estimates of percent canopy cover class and the presence of disease symptoms. Finally, we use a G- test (Sokal and Rohlf 1981) to examine the potential relationship between growth fate over the three year census interval (see Chapter 2) and the various canopy cover classes. This test provides an additional estimate of the relationship between light level and growth.

Measurements of Photosynthetic Rates

A series of field tests are described below that attempt to ascertain the relationship between growth, plant vigor and light in T. taxifolia. Field measurements of photosynthetic rates were used to address the hypothesis that the decline in T. taxifolia was a result of foliar pathogens (Chapter 1). In addition, a

65 series of lab experiments were conducted measuring photosynthetic capacity of T. taxifolia and closely allied species to test hypotheses relating environmental change to plant stress that may have facilitated the decline (Chapter 1). All of these experiments use the LICOR 6200 closed system infra-red portable gas analyzer. The LICOR 6200 gas analyzer is a closed-system device that allows non-destructive sampling of photosynthetic rates by enclosing branchlets of needles into an airtight one liter chamber. The LICOR 6200 measures the rate of CO2 drawdown, along with numerous other parameters (e.g., temperature, light level, and humidity). Instantaneous rates of photosynthesis and conductance are then calculated by standard gas-exchange equations. Photosynthesis, naturally, is dependent on concentrations of CO2 . All experiments were conducted under CO2 concentrations between 350 and 400 ppm (ambient atmospheric conditions).

We began by establishing a baseline photosynthetic light response curve for T. taxifolia. This field test was conducted using first year needles from 15 plants under natural and enhanced light conditions. Light levels were varied from a maximum of 12.2 to 1821 /mols/m2/sec Photosynthetically Active Radiation (PAR). When possible, natural light was used with shade cloth applied to reduce incident light. In many cases, however, natural light was insufficient. In these cases light was enhanced using a 12v tungsten halogen 100 watt projector lamp with reflector (Sylvania EFP). No difference in maximum rates of

66 photosynthesis were observed when plants were illuminated with natural versus artificial light sources. To determine if the presence of symptoms of foliar pathogens are related to carbon gain, maximum net photosynthetic and respiration rates of healthy needles were compared to those bearing needle spots. The difference between the rates of photosynthesis and respiration is a measure of relative carbon gain potential (Fritter and Hay 1981). These measures of differences in rates of photosynthesis between healthy and diseased needles were conducted in the field during the summer of

1991. Measurements were collected between 9:00 AM and 4:00 PM, using either direct solar irradiation, or when necessary, the lamp described above. In all cases, measurements were recorded only when light levels exceeded the light saturation point for T. taxifolia (350 Mmol/m 2 /sec PAR). A comparative study of the photosynthetic response rates across environmental variables for five species within the family

Taxaceae was conducted in order to assess whether T. taxifolia is particularly sensitive to certain types of environmental stress (see Chapter 1). We compared species responses to light level, temperature, humidity, and water stress for the following species: T. taxifolia, T. californica, T, grandis, T. nucifera, and Taxus floridana. The four species in thegenus Torreya were chosen to provide a broad sample of how closely related species respond to environmental conditions. Taxus floridana was chosen because it is related at the family level and is completely

67 overlapping in distribution and habitat with Torreya taxifolia. These tests were performed in the greenhouse and in controlled environment chambers. Seedlings of T. californica and Taxus floridana were 2- to 3-years old and had been grown outside under full sun. The Torreya taxifolia, T. randis, and T. nucifera were derived from cuttings and grown under greenhouse conditions for 2 years. All plants were acclimated to greenhouse conditions for at least one month prior to measurements. All plants were fertilized weekly with 20-20-20 fertilizer and were grown in 5 inch clay plots containing a peat:pearlite:sand (1:1:1 vol.) mixture. In order to maintain similar environmental conditions, and have control over temperature, measurements were conducted in a Conviron growth chamber (Controlled Environments Inc., Pembina ND, USA) with a 12 h photoperiod and day/night temperatures of 220C/180 C. The growth chamber was illuminated by banks of 195 W Sylvania cool white fluorescent bulbs and 40 W General Electric incandescent bulbs. Together these lights provided approximately 400 Mmol/m2/sec PAR at the top of the crown. To insure light saturation during measurements, the 12v 100 W projector lamp described above was used to supplement light within the growth chamber. All measurements were conducted at PAR greater than 500 2 Mmol/m /sec and between 350 and 400 ppm CO2. Temperature response curves of net photosynthesis and dark respiration were conducted at a constant vapor pressure to eliminate the potential effects of varying moisture availability

68 on stomatal aperture (Fritschen and Gay 1979). Measurements of light, and drought photosynthetic response curves were conducted at 200C and 50% relative humidity. The drought study was initiated by saturating pots with water and measuring photosynthesis daily for 6 days. Water was withheld for the duration of the study and the pots were individually weighed just prior to photosynthetic measurements to determine water loss from the pots. Photosynthetic response to humidity was determined by maintaining the appropriate relative humidity (approximately 20%, 50% and 90%) within the leaf chamber at 200C. RESULTS Light and Growth Field tests of light and growth had mixed results. Both maximum and mean terminal extension length were positively correlated with light level (Figure 4.1). In contrast, we found no relationship between the proportion of lateral branch buds that expanded and light (r=.16,p=.62). The observed relationships between terminal extension and light are influenced by the contribution of a single tree with very high light levels. However, the significance of the light and growth relationships remain after exclusion of this outlying observation. One year growth measurements on the 12 trees for which light levels were increased showed no observable increase in growth response. The majority (9 of 12) either died or failed to expand a terminal bud during this first year of the test. Thus, we recorded no positive growth response of trees to increased light

69 length of terminal bud elongation in the past Fiqur 4.1. Average canopy three years of TorreyA taxifolia plotted against percent (see text for explanation of methods). A) light transmission observations without Aspalaga tree (outlier), B) all

A Torreya 16 y = 5.5845 + 0.25749x R^2 * 0.302 14 U

Average 12 Terminal B Shoot 0 Elongation 10

8

6 10 20 30 Percent Canopy Light Transmission

B Torreya

d9b - a& a0% A MaA ^A A^AP^" - iA 1 &% ^LiA 10

16

14 Average Terminal 12 Shoot Elongation 10

8

6 10 20 30 40- 50 Percent Canopy Light Transmission levels treatments. Likewise, canopy cover class was not strongly related to three-year survivorship, although slightly more trees

in high light classes died or suffered major stem death than expected, fewer trees died or suffered major stem death in low light classes than expected, and more trees grew in low light classes than expected (Table 4.1). This result also runs counter to our light limitation hypothesis. Finally, neither the occurrence of needle spots, needle necrosis, or stem cankers appear related to canopy cover class (Table 4.2). Measurements of Photosynthetic Rates A comparison of light curves reveals that the light

saturation point is at approximately 300 Mmol/m 2 /sec PAR for both healthy and diseased needles (Figure 4.2). Healthy needles, however, have an approximately 20 % higher maximum net photosynthetic capability than diseased tissue (Figure 4.3). In addition, the 20% reduction in photosynthetic output of diseased needles exceeds the total surface area effected by the foliar pathogens, indicating a disruption of internal leaf function, rather than a simple reduction in photosynthetically active tissue. Likewise, we measured the dark respiration rates of healthy needles from eight trees versus diseased needles from eight trees. Dark respiration rates were calculated using multiple measurements (1-4) on each tree. Healthy needles have

lower respiration rates diseased needles (Figure 4.3). Measurements of productivity by needle age demonstrates that T. taxifolia retains as much as 90% of maximum net photosynthetic

71 Table 4.1 Variation in survival and growth compared to canopy cover class. Chi-square test of association lumped adjacent cover classes 1, and 2 with 3, 4 with 5, and 6 with 7. Individuals that died were lumped with individuals whose major stem died. Coalescing of groups was done to avoid sensitivity of the analysis as a result of small expected values in accordance with Sokal and Rohlf (1981). Canopy cover class Growth open closed Fate 1 2 3 4 5 6 7 tot Died 0 1 1 0 2 0 0 4 Major Stem Death 0 2 7 5 6 0 0 20 Did Not Grow 0 3 8 8 15 1 1 36 Grew 1 2 3 2 9 3 2 22 Total 1 8 19 15 32 4 3 82 G=13.49, X(.05,7df)=14.07, .05 > p > .1 Table 4.2 Variation in disease symptom by canopy cover class. Canopy Cover Class open closed 1 2 3 4 5 6 7 total Canker Yes 0 3 5 6 14 15 20 63 No 0 6 13 15 29 18 10 91 Total 0 9 18 21 43 33 30 154 Needle Spot Yes 1 3 8 6 16 10 11 55 No 0 6 13 15 30 28 24 116 Total 1 9 21 21 46 38 35 171 Needle Necrosis Yes 0 5 18 13 33 18 21 108 No 1 4 3 9 14 20 14 65 Total 1 9 21 22 47 38 35 173 Needles Chewed Yes .0 3 5 8 11 15 12 54 No 1 3 8 8 12 18 15 65 Total 1 6 13 16 23 33 27 119

All G-tests (within damage category): P>. 25

72 Photosynthesis

2 AMm0lC0 2 /m /2

0 0 0v 0g

0. ~ ENOI I0 01 (Art.( U,. 0 0 0 :0

'7, 0. 0

U'. 0

*m 01' 0~ 0 0 * ~ 0

U' o '-a

t4J U' U~0' 0 0

U' 0

0 0 Ct 1b ~d. ~1 U) 0' Figure 4.3. Comparative rates of A) net photosynthesis and B) dark respiration healthy needles versus bundles of needles in which some needles exhibit needle spotting.

Af 10 -

8- Co 0 LU 0 z 6- c~4 zI- 0' U (0 0 4- 0I. 0 I- 9 0 0 z 0 0. 0 2- S-= 5.56 p<.01 x = 4.42 0 a HEALTHY DISEASED

U Dark Respiration Rate c~4 0 0 t.1.99, pM.07 U A A p.4 U0 -1- I C .2 Cu .5. 0 0 .3soI I Healthy Diseased Needle Condition Figure 4.4. Rates of net photosynthesis for needles of varying age. Measurements collected in A) February, prior to new needle expansion, and B) Autumn.

Maximum Photosynthetic Rates

9

8 7. I 6 e 8 U/) 9 0 C 0 5 S 0 0 0 (0o 0 4. >.0)

0. 3-

1

M . j" II 0 4

Needle Age

September, 1991

(0 5- * *

4. (0

0 o 3"

2 I U _ I - "W- "-It 5 6 0 1 2 3 F Needle Age capacity in fourth year needles (Figure 4.4). Further, this pattern is repeatable throughout the growing season (Figure 4.4). A comparison of greenhouse light response curves among the five species examined here demonstrates that T. taxifolia has a

2 mean maximum net photosynthetic rate (8.3 4mol C02 /m /s) slightly above all but T_ californica (Figure 4.5). In addition, T. taxifolia reaches light saturation at PAR levels approximately equivalent to other closely related species (Figure 4.5).

However, light curves for T. taxifolia from the field and in the greenhouse are different in that maximum net photosynthesis rates, as well as the light saturation point, are higher in greenhouse plants (Figure 4.2, 4.5). Net photosynthesis did not vary between species with respect to relative humidity (Figure 4.6). The suite of species tested all maximized rates of photosynthesis at about 20°C but varied in the shape of their temperature response curves (Figure 4.7). In particular, Taxus floridana and Torreya nucifera showed particularly narrow ranges of temperature over which net photosynthetic rates were high.

Finally, these five species were tested with respect to drought tolerance. These data are presented as mean net photosynthesis and mean stomatal conductance by day because the mean decrease in percent of saturated pot weight did not vary significantly between species (Figure 4.8). In addition, the plants used for this experiment were of equivalent size. Thus, the variable "day" can be used as a proxy for partial water

76 Figure 4.5. Comparative light saturation curves for five closely related species: A) Torreya taxifolia, B) T. califoria )T nucfera,, D) Ta31 floridaia and E) Torreagrandis.

Torrya taxifoll Taxus fioridana 10 0 10' 0. 0.000.0 U) 00 @00.0 * 0. 1 0 0 0 0 0 S6 CM 0 0 04 y 00@0000 0 4. ~00 0 OJO

NoF I W 0 w -- F 0 500 1000- 1500 2000 0 500 1000 1500 2000o

10 1 Torreya callfomlca 10 Torreya grandis 0 0 0 0 *0 0 00 ~ 0 8 0 0 0 Oc *@'0 0 0~0~ 6 0% 0 0 0 '1 .4

Z w I

-, 0' I- v 0 v 0 v I w I-I, i- v 50m w a 5I 1000 1500 2000 1000 1500 2000 PAR (gL morn2 Is)

10- Torroya nucifemr

8' 0 0@ .0 0 .

Ci4 0 0 0* 0

0

I/ - 1 50 0 1000. 1500 2000 PAR (g mor/m2 Is) Figure 4.6. A comparison of net photosyntnetic rates across variation in relative humidity for five closely related species: A) IorrYa taxioir.~Ja, B) TI. californica, C) L6.9.ncifer~a, D) Iax3Ja fJlo.i4dni1 and E) To9rre.ya gani

Torreym taxIfola 11II Taxus fioridana 10 10 9. 9- 6. I 0 I 0 0 7. 7. 0 6. 6.4 5. 5.M 4. 0 = 4.. 3. z ,oo 2.. 2. 1.0 0. mr-1 I * I * I * I * I I I I · I · I 0 20 40 60 80 100 0 20 40 60 s0 100

10 - Torreya cailfomica 102 Torreya grandis ON 9- 9. 8- 7. 7. 0 6- I 62 S 'I 5. 4- 4. 3. 3. 2-. 2. 1I 0. - One 0. U - U - I I w I w I a I 0 2 40 80so 104O 0 20 40 60 80 100 Relative Humidity(% 4.4 lo Torreya nucifera U).. 9. ON. 'C 7. 0 6. i -o0 5. 4. 3. 2. 10 F w 5 I v I * I 1 0 20 40 60 80 100 Relative Humidity (%) Figure 4.7. A comparison of net photosynthetic rates across variation in temperature for five closely related species: A) Tory taxi'folia, B) I. californica, )T nucifera,, D)Tas floridana, and E) Iflreya rra~idis.

Tormya taxifolla Taxus floridana 10

8.

0 I 64 4- E 2,

0l 010 20 30 40 50 0 10 20 30 40 50

1

N

0

E

0 10 20 30 40 50 Temperature ( 0C)

not photosynthesIs E a dark rsplratlonI

010 20 30 40 50 Temperature ( OC) Figure 4.8. A comparison of percent soil moisture through time without water for five closely related species.

100

90

AI SI 80

70) 6?

Uw II .i' r ri 0 1 2 3 4 5 6 7 Day -- o-- Torreya taxifolla --- Torreya califomica =- Torreya grandis * Torreya nucifera -- +-- Taxus 100- floridana

41 80-

60- I. 40- 20" a

An- v -1 I -I I I I I 0 1. 2 3 4 5 6. Day

Figure 4.9. A comparison of mean net photosynthetic rate against time without water for five closely related species. pressure of the soil within pots. These results suggest that net photosynthetic rates in T_. grandis and T_. nucifera are most sensitive to water stress, and that T. taxifolia is as robust, with respect to water stress, as any of the species tested (Figure 4.9).

DISCUSSION The field results presented here provide mixed support for the hypothesis that T. taxifolia is currently stressed by low light. The correlations of growth with light suggest that T. taxifolia thrives in high light situations and that growth is limited by low light. In contrast, preliminary growth response of trees in experimentally created gaps have failed to support the conclusion that low ambient light may be limiting growth in the field. However, we require additional data over time to adequately test whether our experimentally created gaps effect growth in the field. Initial results from a greenhouse growth experiment indicate that light levels below 400 Mmol/m2 /sec PAR significantly limit growth rate (Schwartz et al. unpublished data). The low light saturation point (350 Mmol/m2 /sec PAR) of T. taxifolia and the frequency with which they are found under dense shade (Chapter 2) suggests that the tree can remain suppressed under fairly dense shade for extended periods. An analysis of tree-ring cores from long-dead trees would help clarify this question. A cursory examination of tree cores from dead logs appear to exhibit groups of wide and narrow annual growth rings characteristic suppression and release growth patterns (Schwartz

81 personal observation). At the present time, most of the trees in the census population are found in light environments far below the light saturation point (Chapter 2, p32). Finally, the results from the three-year census indicate little relationship between light level and survivorship. The data we present on differences in net photosynthetic rates between healthy and diseased tissue, as well as needles of differing ages, indicate that a foliar pathogen remains a plausible candidate for the cause of the decline. However, one must bear in mind that these results show a correlation between disease and reduced net photosynthesis and do not prove cause and effect. Although, the fact that we often measured healthy needles V immediately adjacent to diseased needles indicates a highly localized infection, typical of foliar pathogens. Foliar pathogens have typically been considered unlikely candidates for catastrophic declines because they rarely kill the trees that they infect (Manion 1981). Foliar pathogens can defoliate trees in a single season,. but this is rarely sufficient to kill a tree. However, T.taxifolia is different for several reasons. First, we have shown that needles that bear pathogens have a reduced capacity for carbon gain, a necessary but insufficient pattern, if we are to believe that foliar pathogens are responsible for this decline. This reduction in net photosynthesis will result in plant' stress if, as we suggest, the pathogen causes the decreased physiological activity, and is not merely a secondary effect. Second, most conifers demonstrate a

82 marked decline in photosynthesis with needle age. In contrast, needles up to four years of age on T. taxifolia are photosynthetically active. Thus, defoliation of older needles, as is often observed, would significantly reduce the total photosynthetic capacity of the plant. Further, these older needles are not readily replaced. Each year's single flush of needles represent a small component of the plant's total photosynthetic tissue. The observation that only about 70 * of lateral branch buds and less than 30% of terminal branch buds elongate each year (Chapter 2) indicates that it may take several years for a defoliated tree to recover a full compliment of green tissue. Since T. taxifolia is most frequently found as small suppressed trees beneath a dense forest canopy, most trees are probably close to the minimum carbon gain levels for survival. The overstory of many of these ravine forests has been in a maturation cycle for the past 30-50 years since selective cutting of hardwoods in the mid-1900's. In a maturing forest there are likely to be relatively few canopy gaps providing high light levels for release of suppressed trees. The additional stress of a single bout of heavy defoliation may have been sufficient to kill most trees. If these factors are not fully responsible for the decline, then they are at least likely to help explain why T., taxifolia has not recovered from the decline of 30 years ago. This scenario, however, remains speculative until a foliar pathogen can be isolated and used to test the foliar pathogen hypothesis.

83 Our comparison of temperature, relative humidity and drought response of T. taxifolia to other closely related species reveals that we have no reason to suspect that T. taxifolia should be more susceptible to environmental stresses than other similar species. This is particularly interesting in light of the fact that Taxus floridana grows over the same range and in the same habitat as T. taxifolia, yet was not affected by a decline in population size. Further, the similar response between T. taxifolia and T. californica with respect to water stress is somewhat surprising because the climate in the habitat of T. californica is Mediterranean, and very drought prone. Finally, T. taxifolia appears to be more tolerant to short-term water stress than Taxus floridana. We would, again, predict the opposite if T. taxifolia became susceptible to disease as a result of drought stress, as has been suggested (USFWS 1986, Chapter 1). Thus, we find no support for the hypothesis that environmental stress, namely increased ravine temperature, decreased humidity, or decreased soil moisture, would have rendered T. taxifolia particularly susceptible to natural pathogens so as to cause the catastrophic decline. While all of these stresses may contribute to limit carbon gain, they do not seem to effect T. taxifolia to a greater extent than other species, particularly a closely related potential competitor, Taxus floridana.

84 CHAPTER 5

The Foliar Fungal Associates of Torreya taxifolia:

Pathogenicity and Sensitivity to Smoke

As stated previously (Chapter 1), there is no apparent virulent introduced pathogen involved in the decline Torreya taxifolia. The root of the problem in isolating potential environmental mechanisms for the cause of the decline is in isolating a pathogen. Here we present data that describe our attempts to isolate pathogens from T_.taxifolia and establish pathogenicity. However, all these attempts at establishing pathogenicity of fungal associates have failed. We have not been able to determine the cause, proximate or ultimate, for the decline of T. taxifolia. Unfortunately, these pathogenicity tests also do not exclude the potential pathogens that we tested from further consideration. We have isolated several potential pathogens, and tested them for pathogenicity. However, we cannot say that these fungi might not behave as pathogens under a particular set of environmental conditions that we have not tested.

We have also been testing certain foliar fungal associates for susceptibility to smoke. Through these tests we can assess the likelihood of the fire suppression hypothesis for the decline of T. taxifolia (see Chapter 1). This hypothesis suggests that

85 smoke from frequent fires in the adjacent uplands maintained low populations of potential pathogens. Fire suppression, in the 1950's, allowed pathogen populations to explode, precipitating a catastrophic population decline. Fungal associate isolation strategies, pathogenicity tests, and smoke susceptibility tests are described below. METHODS Fungal Isolation All needles collected for fungal isolation were from wild- grown trees. Attempts to isolate fungal pathogens from needle and stem tissue have been repeated in all seasons over the past three years. We used needles that exhibited foliar symptoms of disease, such as needle spots and needle necrosis, as well as those that appeared healthy. We have sampled needles from trees throughout the range of the species. Needles of T. taxifolia collected for fungal isolation were surface sterilized in a 10% bleach solution, a 50% alcohol solution, a flame sterilization, or any, and all, combinations of these techniques. Needles were immersed in sterilizing solution for 1-to 3 minutes. Needles were either cut or placed on artificial medium whole. Stem tissue was also collected from wood tissue and surface sterilized in the same manner as needles. Wood samples were healthy portions of cambium or heartwood adjacent to cankers. We used several types of artificial growth medium for our isolation attempts. These

include: Potato Dextrose Agar (PDA), Malt Agar, Water Agar, Corn Meal Agar, and Water Agar with sterilized straw. By using a

86 variety of techniques we hoped to isolate as many different potential pathogens as possible. Pathogenicity Tests A selection of foliar fungal isolates, derived from the aforementioned methods, were tested for their pathogenicity on T. taxifolia. A fungal spore slurry of at least 100,000 spores per cc was misted onto test plants in a controlled environment chamber. Plants used for pathogenicity tests were divided into two groups. One group received full water treatment, while the other was maintained under water stress conditions. Water stress was defined as a soil moisture of between 70 and 80 percent of saturated pot weight. This level of soil moisture was previously observed to result in a significant decrease in photosynthesis (see Chapter 4). Plants were maintained in these treatments for a minimum of two weeks after inoculation. Fungal Associates Susceptibility to Smoke Smoke susceptibility in fungal isolates was tested in three experiments. The first experiment tests the effect of smoke residue on fungal colony growth. Clean culture plates were placed, top open, in a closed chamber and subjected to smoke for varying lengths of time from 0 to 20 minutes. Five replicate plates for each smoke treatment were then inoculated with a small fungal colony, not exceeding mm in diameter.. All plates were placed at room temperature and illuminated by sunlight for the duration of one week. Colony diameter was measured on a daily basis throughout the experiment. The experiment ended when the

87 unsmoked plates reached the edge of the culture plate, in approximately one week.

The second experiment tested the direct effect of smoke on fungal colony growth. Colonies were begun in replicate culture plates and allowed to grow to approximately 2 cm diameter. Five replicate plates with fungal colonies were then smoked as per the procedure described above. Subsequent colony growth was recorded as above. Smoke for these treatments was cooled along a 20 ft length of 6 inch duct pipe. The pipe was cooled by running water. Thus, we eliminated any direct effect of high temperature on the fungal colonies during smoke treatment.

A final experiment tested the effect of smoke residue on fungal spore germination. We used five replicate smoked culture plates for three time treatments (0, 1, and 5 minutes), as in the first experiment. After smoke treatment a slurry of fungal spores was spread over the smoked culture medium on each plate. Spores were examined after 12 hours to-determine rates of spore germination.

For each experiment clean glass slides were placed on the tray at the time of smoke treatment. The glass slides were scanned in a spectrophotometer (Bausch and Lomb Spectronic 70) at 470 nm to measure percent light transmission after smoke deposition. Percent light transmission was measured for each smoke treatment level as determined by time. A linear °i relationship exists between percent light transmission and time of smoke treatment (Figure 5.1).

88 .Figure 5.1. Biplot of the relationship between time in smoke chamber and density of smoke residue deposited on clean slides in chamber. glass Smoke residue is measured as percent of light transmission through a clean glass slide'at 470nm.

100

90 8 80 0 ZI E 70 8

cu 60-

50 10 20 30 Time (Minutes)

Table 5.1 Fungi isolated from Torreya a ~iQ1± in the native range. species Smoke 1st iuoo 1st spore £lUAr.23 pe x x kcrimnamsp. x- x ------A mixture of pine straw and wiregrass was collected from mature upland longleaf pine forests within the natural range of Torreya taxifolia for use as fuels for all experiments. These fuels are representative of the fuels in natural and managed

fire. We chose the most frequently isolated fungi in our tests of smoke susceptibility. The fungal species used in each these experiments are listed in Table 5.1.

Field Tests of Smoke Hypothesis A field test of the amount of smoke deposited into ravines through upland fire management was begun in the spring of 1992.

This experiment was delayed as a result of a poor fire season in 1991. Thus, the results from this experiment constitute smoke from only a single fire treatment, and must be regarded as preliminary.

Finally, a field test of the smoke hypothesis was begun in 1989. Twenty pairs of trees have been planted in the wild for testing. One tree from each of these pairs was to have been smoked each spring. We would then monitor the incidence of new disease in these trees. As a result of vandalism this experiment has been temporarily abandoned. We have lost 7 of our 40 plants to natural causes that appear to be related to transplant shock.

None of these plants exhibited symptoms of disease. An additional 17 plants have been dug up over the two year period of this test. We recently discovered the culprit and this disruption will now cease. However, we have lost the bulk of our data from this experiment through this vandalism. The remaining plants, however,

90 appear healthy and show no symptoms of new disease.

RESULTS

Fungal Isolation and Pathogenicity During the course of three years we have isolated three new potential pathogens from T. taxifolia (Table 5.1). Pestalotia natens is an exceedingly common fungus found on nearly every T. taxifolia throughout the natural range. This fungus is readily isolated using PDA. Fusarium sp. was also frequently isolated on PDA from both needles and cambium tissue. A species from the genus Acremonium was infrequently isolated on a malt extract agar. Finally, Botrysphaerium sp. was isolated on corn meal agar. These four isolates were tested for pathogenicity on both well watered and moisture stressed plants. All tests were negative. Our list of fungal isolates varies somewhat from previous studies (Alfieri et al 1967, USFWS 1986). The explanation for this appears to lie in the fact that we used native trees for our experiments, while previous studies used trees growing in the wild as well as trees in lawns, at Maclay Gardens and on the campus of the University of Florida (Alfieri et al 1967, El-Gholl personal communication). These trees outside the natural range are likely to be vectors of other horticultural diseases that may not infect wild trees. Thus, we prefer our list of fungal isolates with one exception. Phyllosticta sp. was isolated from ascospores on needles in the wild (Alfieri et al 1967). We did not focus on this species because we could not find the symptomatic ascospores on trees in the field as described by El

91 Gholl (personal communication), and because it was tested for pathogenicity before with negative results (El-Gholl 1984). Smoke Susceptibility The four fungal isolates used for pathogenicity tests were also tested for susceptibility to smoke. The results are mixed.

Fusarium sp. and Pestalotia natens are both highly susceptible to smoke and show decreased fungal growth in response to direct smoke treatment (Figure 5.2), as well as smoke residue on previously smoked plates (Figure 5.3). In addition, both of these species show reduced spore germination (Figure 5.4), reduced spore concentration (Figure 5.5) and reduced growth in germinating spores (Figure 5.6) in response to smoke residue on culture plates. In contrast, Acremonium and Botrysphaerium demonstrate virtually no response to any level or type of smoke treatment (Figure 5.2, 5.3). Field tests of the ability of smoke to inhibit disease in T. taxifolia have been unfruitful. To date, none of the plants that remain in our experiment, smoked or unsmoked, shows any characteristic symptoms of disease (needle spots, needle necrosis, or stem cankers). In addition, we monitored one managed fire for smoke deposition in ravines. Smoke residue (collected on glass slides) exceeded the level expected from our minimum experimental smoke treatment (95% transmission at 470nm for 1 minute smoke treatment - Figure 5.1) in only 1 of 48 samples.

This single sample was at the highest slope position and adjacent to the fire line.

92 Figure 5.2. The effects of direct smoke treatment on growth of fungal associates of L. taxiolUi. A) Pestalotia natans, B) Acremonium sp. Each point is .the mean of growth observed in five replicate petri plates containing PDA agar and stored at room temperature.

8 a

6· 0 min 1 min E -U-- 3 min 5 min U 10 min .5a 20 min 30 min

w 1' ' I" 0 1 ~2 3 4 5 day

8 rý

U 0 min p 1 min 3 min I0 U 5 min .2C 2 10 min 0 20 min U a*. iwLil -I I - - ommmmrý 0 7 day Iiit.I

3..f

0*.0I

W0 .4 6 4J 0 A E 0 S U ph, o t U .4 0 41 0 0 0 o u -P4 LL

0a 0 0 C 0 ml a a W4J $WW4 0

>44)

*Adcor~. 0 cccC 'I I l IEI E 0?10Y Eu4-- > t 0 X - W~ M au 4)Fa. I . I

0o

Ur

0r

0

(wa) *wLipAuoioo (us*)welp)Auol(* Figure 5.4. Histograms of spore germination failure rate for various smoke treatments (0, 1, 3 and 5 minutes). Replicate sterile PDA agar filled petri plates were inoculated with a slurry of spores from A) Fusarium sp. and B) Pestalotia natans. Germination rates were observed following 16 hours of incubation at room temperature.

A.) Fusarium 40

30 Germination Failure 20 (%)

10

0 0 1 3 5 Smoke Treatment (minutes)

B.) Pestalotia 20

Germination Failure lO (%)

0 0 1 3 5 Smoke Treatment (minutes) Figure 5.5. Histograms of spore density for various smoke treatments (0, 1, 3 and 5 minutes). Replicate sterile PDA agar filled petri plates were inoculated with a slurry of spores from A) Fusarium sp. and B) Pestalotia natans. Spore density, reflecting differences in rates of spore deterioration, were observed following 16 hours of incubation at room temperature.

A.) Fusarlum 300

Spore 200 Density 13 fde (#/tran.) 100

0 0 1 3 5 Smoke Treatment (minutes)

B.) Pestalotia 1

Spore Density (#/tran.)

0 1 3 5 Smoke Treatment (minutes) Figure 5.6. Histograms of me4n spore growth rates for various smoke treatments (0, 1, 3 and 5 minutes). Replicate sterile PDA agar filled petri plates were inoculated with a slurry of spores from A) Fusarium sp. and B) Pestalotia natans. Spore growth, as measured by the length of new mycelial strands were observed following 16 hours of incubation at room temperature.

A.) Fusarium sp. 300 n,,20I n-20UI

200- Growth (microns) n=20 n=20 100- n=20

O0I 0 3 5 Smoke Treatment (minutes)

B.) Pestalotia sp.

Growth (microns

0 1 3 5 Smoke Treatment (minutes)

5 DISCUSSION The results of our pathogenicity tests are not encouraging. The failure to confirm pathogenicity in any fungal isolate could either mean that we have failed to isolate the disease agent, or that the conditions under which the disease agent becomes pathogenic are restrictive and were not encountered during our tests. The fact that none of our experimental plants in the field have exhibited symptoms of disease supports the contention that the foliar fungi only become pathogenic under severe plant stress. Although this lack of disease symptoms in field tests plants.is also consistent with another hypotheses: the disease agent that decimated T. taxifolia has been reduced or eliminated from the natural population and thus most T. taxifolia no longer encounter the disease agent (see Chapter 1). Alternatively, the disease agent that caused the population decline may still infect plants in the wild. If the pathogen requires the host plant to survive, such as most rust fungi, we would not be able to isolate it on artificial growth medium. Thus, we would get no positive pathogenicity tests using the fungi we did isolate. If this is the case, the isolates from this study may represent primarily secondary pathogens of little consequence to plant health. However, it is hopeful to note that the test plants we used were derived from cuttings of diseased trees. If a host-specific pathogen is involved, it is likely to be systemic and would have been transferred to cuttings collected from the wild (for propagation for ex situ conservation).

98 However, these cuttings do not exhibit the symptoms of disease observed in the wild. In addition, systemic rust fungi produce. fruiting bodies on the outside of host plants. We have searched plants in the wild for such symptoms of systemic infection. While several fruiting bodies on T. taxifolia were observed in the course of this work, most were identified as saprophytes that appear to be derived externally, using the trunk for support of its fruiting body structure (L. Crane, pers. comm.). Collectively, these results suggest that the disease that decimated T_.taxifolia may no longer be found. This hypothesis requires that disease symptoms currently observed are secondary pathogens breaking out on plants stressed by low light, low nutrients or low soil moisture. While most foliar pathogens are considered secondary infections, our evidence neither strongly supports, nor rejects, this hypothesis. Our test of the smoke hypothesis yielded mixed results. Two of the four fungal associates tested proved sensitive to smoke while the other two are not. Thus, it is possible that smoke played a role in reducing foliar pathogen attack, but only if the attacking agent is a fungus that is susceptible to smoke. Several other, as yet untested, features must also hold for the smoke hypothesis to be supported. It remains to be satisfactorily tested whether smoke from managed fires settles in ravines in sufficient quantities to maintain reduced populations of potential pathogens. Further, we do not yet know the frequency of

fire required to maintain reduced fungal populations.

99 CHAPTER 6. General Conclusions and Recommendations

In many respects the results of this 'research are exactly what we would have predicted. First, the species may be in decline for any number of reasons, a subset of which were' previously reported in the US Fish and Wildlife Service Recovery Plan (USFWS 1986, Chapter 1). Second, the species is continuing to decline, as expected of plants with no sexual reproduction that are also being suppressed beneath a relatively dense, young closed canopy. Given that the species is exceedingly rare and that the decline eliminated all mature seed producing trees, this decline seems, if anything, remarkably slow (Chapter 2). Third, the species is described by a low amount of genetic variability

(Chapter 3). This is consistent with the pattern of several ancient species that have become narrowly endemic (Chapter 3).0 Fourth, the photosynthetic physiology of the species confirms-that canopy shading could be suppressing the recovery of what appear to be otherwise potentially healthy individuals (Chapter 4). Fifth, the photosynthetic response of the species to water and temperature stress does not indicate any special consideration to these factors as potential contributors to the demise of the species (Chapter 4). Sixth, the foliar pathogens

100 are associated with a significant decline in photosynthetic capacity of T. taxifolia, consistent with the hypothesis that foliar pathogens could become primary disease agents under appropriate conditions (Chapter 4).

Seventh, we failed to contribute any significant advancement in the realm of discerning a disease agent (Chapter 5). Finally, smoke treatment may be important to the health of T. taxifolia. This, however, depends on two specific factors that are yet to be established: 1) the disease agent must be one of the foliar fungi that is susceptible to smoke (although we have shown that these exist), 2) the amount and frequency of smoke treatment required to keep populations of the potential pathogen at low levels must be on the order of smoke intrusion into ravines under natural fire conditions.

We recommend that future research on the conservation program for Torreya taxifolia should proceed along four lines. First, continue efforts to isolate a primary pathogen, particularly with respect to host-specific rust fungi. Second, continue experiments that provide canopy openings to further test if these facilitate recovery of vigorous individuals in the wild. Third, attempt test plantings of closely related species (in particular Torreya californica) to assess whether these species are vulnerable to the diseases of T. taxifolia. This recommendation is particularly important because of the ex situ conservation program. We need to establish whether we would jeopardize T. californica by incidental contact with T.

101 taxifolia. This could be an issue because the US Forest Service

Tree Genetics lab, perhaps the most capable institute to conduct tissue culture propagation of T. taxifolia in the event that systemic pathogens are found, is located near the range of T. californica. As a result of this close proximity with the range of T. californica, this lab has been heretofore excluded from housing ex situ specimens of T. taxifolia. Finally, we recommend additional test plantings to determine how newly planted trees respond with respect to new disease and growth. In addition, we make the following management recommendations. First, monitoring of large populations should continue on an annual basis. Second, canopy gaps should be maintained around several of the larger, healthy trees in an effort to stimulate sexual maturity, particularly among the few trees that seem like they may be female. Third, several deer exclosures were erected in 1987 to protect large stems from deer antler rubs. These have largely failed. Those that remain may be removed and no further exclosures need be attempted at this time. While the exclosures have succeeded by preventing further antler rub damage, they increase the sphere of influence of woody debris falling from the canopy. Several trees have been damaged as a result of limbs hitting a deer exclosure cage and dragging it over the top of the tree it was meant to protect. Thus, it is not clear which is the lesser of these two problems. However, on-going stocking of young pines in the adjacent uplands around trees on preserves in the

102 southern end of the range may help alleviate the pressure for antler rub resources. Deer antler rubs have not appeared to be as important in more northerly populations. Fourth, large-scale restocking of the population with individuals propagated from cuttings should be delayed until more is known about the pathogens. Finally, fire management procedures for uplands adjacent to ravines containing populations of T. taxifolia should be adjusted in an attempt to maximize smoke deposition to ravines under the constraint of fire management safety guidelines. The prospects for recovery of T. taxifolia in the near future remain dim. However, with the current coordinated efforts for ex situ propagation the species is unlikely to suffer extinction. It is hoped that research beyond this project will eventually make in situ recovery of T. taxifolia a possibility.

103 LITERATURE CITED

Alfieri, S.A.Jr., A.P. Martinez, and C. Wehlburg. 1967. Stem and needle blight of Florida torreya, Torreya taxifolia Arn. Proceedings Florida State Horticultural Society 80:428-431. Alfieri, S.A.Jr, K.R. Langdon, C. Wehlburg, and J.W. Kimbrough. 1984. Index of plant diseases in Florida. Florida Department of Agriculture and Consumer Services, Bull. 11.

Ayres, P.G. 1978. Water relations of diseased plants. pp. 1- 60 in: Water Deficits and Plant Growth Vol 5. T.T. Kozlowski, Ed. Academic Press, New York.

Baker, W. 1982. A distribution survey of Taxus floridana and Torreya taxifolia in Florida. Unpubl. report to the U.S. Fish and Wildlife Service.

Baker, W and Leonard. 1982. A distribution survey of Taxus floridana and Torreya taxifolia in Florida. Unpubl. report to the U.S. Fish and Wildlife Service.

Barnes, L.R. 1985. Clonal propagation of endangered native plants Rhododendron chapmanii Gray, Taxus floridana Nutt., and Torreya taxifolia Arn. unpublished Ph.d. dissertation, Univ. of Florida, Gainesville.

Bormann, P.H. 1990. Air pollutionand temperate forests with creeping degradation. In The Earth in Transition: Patterns and Processes of Biotic Impoverishment, G.M. Woodwell, Ed. Cambridge Univ. Press, Cambridge. pp 25-44.

Braun, E.L. 1950. Deciduous Forests of Eastern North America. Blakiston Co., Philadelphia. 596pp.

Burdon, J.J. 1987. Disease and Plant Population Biology. Cambridge Univ. Press, Cambridge. 208p.

Burke, J.G. 1978. Human use of the California nutmeg tree, Torreya californica, and other members of the genus. Economic Botany 29:127-139.

Chapman, A.W. 1885. Torreya taxifolia Arn., a reminiscence. Botanical Gazette 10:251-254.

Conkle, M.T., P.D. Hodgkiss, L.B. Nunnally, and S.C. Hunter. 1982. Starch gel electrophoresis of conifer seeds: a laboratory manual. USDA Forest Service Gen. Tech. Rep. PSW-64. pp. 1-18. Cook, R.J. 1973. Influence of low plant water and soil potential on disease caused by soil borne fungi. Phytopathology 63:451-458.

104 critchfield, W.B. 1984. Impact of the Pleistocene on the genetic structure of North American conifers. In Proc. 8th North American Forest Biology Workshop, ed. R.M. Lanner, pp. 70-118. Logan, Utah. Davy, A.J. and R.L. Jeffries. 1981. Approaches to the monitoring or rare plant populations. Pp. 219-232 in H. Synge, ed., The Biological Aspects of Rare Plant Conservation. Wiley, Cambridge. E1-Gholl, N.E. 1985. Fusarium lateritium causing needle spots on Torreya taxifolia in Florida. Plant Disease 69:905.

Falk, D.A. and K.E. Holsinger. 1991. Genetics and Conservation of Rare Plants. Oxford University Press, New York. 283p. Fritschen, L.J. and L.W. Gay. 1979. Environmental Instrumentation (ppl23-129). Springer-Verlag. Godfrey, R.K. 1988. Trees. Shrubs. and Woody Vines of Northern Florida and Adjacent Georgia and Alabama. Univ. of Georgia Press, Athens, 734p. Godfrey, R.K. and H. Kurz. 1962. The Florida torreya destined for extinction. Science 136:900-902.

Gray, A. 1889. A pilgrimage to Torreya. In: Scientific Papers of Asa GrAy Volume II. selected by C.S. Sargent. Houghton Mifflin Co. New York.

Griffin, D.M. 1978. Effect of soil moisture on survival and spread of pathogens. pp 175-197 in: Water Deficits and Plant Growth Vol 5. T.T. Kozlowski, Ed. Academic Press, New York. Hamrick, J.L. and M.J.W. Godt. 1989. Allozyme diversity in plant species. In Plant Population Genetics, Breeding, and Genetic Resources, ed. A.H.D. Brown, M.T. Clegg,A.L. Kahler and B.S. Weir. pp. 43-63. Sinauer, Sunderland, MA. Harper R.M. 1914. Geography and vegetation of northern Florida. Florida Geol. Survey Sixth Ann Report. 451p.

Harper, F. and A.N. Leeds. 1937. A supplementary chapter on Franklinia alatamaha. Bartonia 19:1-13.

Hepting, G.H. 1963. Climate and forest diseases. Annual Review Plant Physiology. 15:363-386.

Komarek, E.V., Sr. 1964. The natural history of lightning. Proc. Ann. Tall Timbers Fire Ecol. Conf. 3:139-183. Komarek, E.V.,Sr. 1968. Lightning and lightning forces as ecological forces. Proc. Ann. Tall Timbers Fire EQcol. Conf. 105 8:169-197.

Kurz, H. 1938a. Torreya west of the Apalachicola River. Proc. Fla. Acad. ci. 3:66-77. Kurz, H. 1938b. A physiographic study of the tree associations of the Apalachicola River. Proc. Fla. Acad. Sci. 3:78-90. Ledig, F.T. and M.T. Conkle. 1983. Gene diversity and genetic structure in a narrow endemic, Torrey pine (Pinus torreyana Parry ex carr.) Evolution 37:79-86.

Manion, P.D. 1981. Tree Disease Concepts. Prentice Hall, 399p. Melching, J.S., J.R. Stanton, and D.L. Koogle. 1974. Deleterious effects of tobacco smoke on germination and infectivity of spores of Puccinia graminis tritici and on germination of spores of Puccinia striiformis, Pyricularia oryzae and an Alternaria species. Phytopathology 64:1143- 1147.

Melillo, J.M., T.V. Callaghan, F.I. Woodward, E. Salati and S.K. Sinha. 1990. Effects on ecosystems. In: Climate Change: The IPCC Scientific Assessment, J.T. Houghton, G.J. Jenkins and J.J. Ephraums, eds. Cambridge Univ. Press, Cambridge. pp 283-310. Mihail, J.D. 1979. Toxic effects of straw smoke on bean rust, snapdragon rust and on seed germination. Unpubl. M.S. thesis, Univ. California, Berkeley. Millar, C.I., S.H. Strauss, M.T. Conkle, and R.D. Westfall. 1988. Allozyme differentiation and biosystematics of the Californian closed-cone pines (Pinus subsect. Oocarpe). Syst. Bot. 13:351- 370.

Moran, G.F., J.C. Bell, and K.C. Eldridge. 1988. The genetic structure and conservation of the five natural populations of Pinus radiata. Can . For Res. 18:506-514. Nei, M. 1977. F-statistics and analysis of gene diversity in subdivided populations. Ann Human Genet. 41:225-233.

Nei, M. 1978. Estimation of average heterozygosity and genetic distance from a small number of individuals. Genetics 89:583-590. Nelson, R.M. 1932. Growth and mortality of chestnut sprouts. J of Forestry 30:872-873.

Parker, J. 1965. Drought resistance in woody plants. Botanical Review 22:241-289.

Parmeter, J.R.,Jr. and B. Uhrenholt. 1975a. Effects of smoke on pathogens and other fungi. Proc. Ann. Tall Timbers Fire Ecol. 106 Conf. 14:299-304.

Parmeter, J.R.,Jr. and B. Uhrenholt. 1975b. Some effects of pine- needle or grass smoke on fungi. Phytopathology 65:28-31.

Platt, W.J. and M.W. Schwartz. 1990. Temperate hardwood forests. Pages 194-229 in Ecosystems of Florida (R.L. Myers and J.J. Ewel, eds). Univ. of Central Florida Press, Orlando.

Pratt, B.H., W.A. Heather, and C.J. Shepherd. 1973. Origin and distribution of Phytophthora cinnamomi Rand. in Australian native plant communities and significance of its association with particular plant species. Aust. J. Biology 26:559-573. Pyne, S.J. 1982. ire in America: A Cultural History of Wildland and Rural Fire. Princeton Univ. Press, Princeton, New Jersey. Reinsmith, W.H. 1934. Exploring for torreya trees in the Apalachicola bluff country. Unpubl. report. Florida Forest Service, Tallahassee. Robbins, L.E. and R.L. Myers. 1989. Seasonal effects of prescribed burning in Florida: a review. Report prepared for the Florida Game and Freshwater Fish Commission. Savage,. T. 1983. A Georgia station for Torreya taxifolia Arn. survives. Florida Scientist 46:62-64. Schoeneweiss, D.E. 1975. Predisposition, stress, and plant disease. Annual Review Phytopathology 13:193-211. Schoeneweiss, D.E. 1978. Water stress as a predisposing factor in plant disease. pp. 61-99 in: Water Deficits and Plant Growth Vol 5. T.T. Kozlowski, Ed. Academic Press, New York. Schwartz, M.W. 1990. Conserving forest diversity in northern Florida: From landscapes to populations. Phd dissertation. Florida State University, Tallahassee, 170p. Sneath, P.H.A., and R.R. Sokal. 1973. Numerical Taxonomy. W.H. Freeman, San Francisco. Sokal, R.R. and F.J. Rohlf. 1981. Biometry, Second Edition. W.H. Freeman and Co. New York. Stalter, R. and 8. Dial. 1984. Environmental status of the stinking cedar, Torreya taxifolia. Bartonia. 50:40-42. Struhsaker, T.T, J.M. Kasenene, J.C. Gaither, Jr. N. Larsen, 8. Musango and R. Bancroft. 1989. Tree mortality in the Kibale Forest, Uganda: a case study of dieback in a tropical rain forest adjacent to exotic conifer plantations. Forest Egol. and Management 29:165-185.

107 Swofford, D.L. and R.B. Selander. 1989. BIOSYS-1: A computer program for the analysis of allelic variation in population genetics and biochemical systematics. Release 1.7. Illinois Natural History Survey, Champaign, IL. Tebeau, C.W. 1980. A History of Florida 2nd ed. University of Miami Press, Coral Gables, Florida. 527p.

Toops, C. 1981. The 'stinking cedar' is in big trouble. American Forests 87:46-51.

Travis, J. and R. Butter. 1986. Experimental designs and statistical methods for demographic studies of rare plants. Nat. Areas J. 6:3-12. U.S. Fish and Wildlife Service. 1986. Florida Torreya Torreya taxifolia Recovery Plan. U.S. Fish and Wildlife Service, Atlanta, Georgia. 42pp. Warrington, S. and J.B. Whittaker. 1990. Interactions between sitka spruce, the green spruce aphid, sulphur dioxide pollution and drought. Environmental Pollution 65:363-370. Woods, F.W. and R.E. Shanks. 1959. Natural Replacement of chestnut by other species in the Great Smoky Mountains National Park. Ecology 40:349-361.

Woodvell, G.M. 1990.eTh Earth in Transition: Patterns and Processes of Biotic Impoverishment. Cambridge Univ. Press, Cambridge, 530p. Yarvood, C.E. 1959. Predisposition. in Plant Pathology. J.G. Horsfall, A.E. Dimond, eds. Academic Press. New York. Zagory, D. and J.R. Parmeter, Jr. 1984. Fungitoxicity of smoke. Phytopathology. 74:1027-1031.

108 This page is intentionally blank.