ECOLOGICAL AND DEM OGRAPHIC TRENDS AND PATTERNS OF Metamasius callizona (CHEVROLAT), AN INVASIVE BROMELIAD-EATING WEEVIL, AND FLORIDAâ€™S NATIVE BROMELIADS By TERESA M. COOPER A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2006
Copyright 2006 by Teresa M. Cooper
iv ACKNOWLEDGMENTS I thank my major professor, Dr. Howard Frank, for his guidance, support, and patience. I thank my committee members, Dr . Ron Cave, for his many hours of help in the field and for sharing his knowledge, and Dr. Emilio Bruna, for helping me with data analysis. It has been a privilege a nd a pleasure working with my committee. I thank Dr. Barbra Larson who was always so competent and helpful, in a thousand ways. I thank Dr. Ken Portier for he lping me with statistical analysis. I thank the Florida Park Service for funding my research. I thank biologists Paula Benshoff, Marian Bailey, Dorothy Harris, Mike Owen, and Jill Scanlon for introducing me to their respective parks, preserves, and refuge, for helping me whenever I needed their help, and for accommodating volunteer participation. I thank the volunteers, Kathy Walters, Berni Reeves, Ruth Slayter, Lucille Weinstat, Ron Fleck, Cathy Bergens, Tom F unari, Huchiro and Susan Shimanuki, John Roman, Karen Relish, Patrick Duetting, and Ce lia Branch, for their cheerful dedication. I thank my family for believing in me and for being so understanding. I thank Dr. Richard Freed for listeni ng so kindly, for offering continual encouragement, and for being such a good friend. I thank Dr. Alvin Lawrence for being so deft in handling porcupines. I thank Yunit Armengol for granting me inspiration. And I thank Jimmy Yawn for his love, his patience, his photography, and his outstanding talents as camp cook.
v TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES............................................................................................................vii LIST OF FIGURES.........................................................................................................viii ABSTRACT....................................................................................................................... ..x CHAPTER 1 INTRODUCTION........................................................................................................1 2 METHOD AND MATERIALS..................................................................................10 3 DOES Metamasius callizona (CHEVROLAT) EXPERIENCE SEASONAL POPULATION FLUCTUATIONS?..........................................................................16 Introduction.................................................................................................................16 Method and Materials.................................................................................................16 Results and Discussion...............................................................................................17 4 SEASONAL TRENDS AND PATTERNS OF Tillandsia fasciculata SWARTZ AND Tillandsia utriculata L. POPULATIONS IN MYAKKA RIVER STATE PARK..........................................................................................................................2 2 Introduction.................................................................................................................22 Methods and Materials...............................................................................................23 Results and Discussion...............................................................................................26 5 SURVIVAL OF Tillansia fasciculata SWARTZ AND Tillandsia utriculata L. IN MYAKKA RIVER STATE PARK............................................................................35 Introduction.................................................................................................................35 Method and Materials.................................................................................................36 Results and Discussion...............................................................................................38 6 Metamasius callizona (CHEVROLAT) AND THE FUTURE OF FLORIDAâ€™S NATIVE BROMELIADS..........................................................................................48
vi Introduction.................................................................................................................48 Methods and Materials...............................................................................................49 Results and Discussion...............................................................................................50 APPENDIX A METHOD TIERS.......................................................................................................62 B MAPPING SCHEDULE............................................................................................63 C LONGEST LEAF LENGTH......................................................................................64 D HEALTH RATINGS..................................................................................................65 LIST OF REFERENCES...................................................................................................66 BIOGRAPHICAL SKETCH.............................................................................................69
vii LIST OF TABLES Table page 4-1 Cross correlations for compari ng the three bromeliad data sets..............................34 5-1 Test statistics for null hypothesis that all three survival curves are the same for small, medium and large class-size bromeliads.......................................................44 5-2 Test statistics for null hypothesis that the survival curves are the same for two modes of death (weevil-kill or other-kill) for large class-size T. fasciculata ..........46 5-3 Median values (months) for failure m odes other deaths and disappearances vs. weevil deaths for large class-size T. fasciculata , MRSP..........................................46 5-4 Test statistics for null hypothesis that the survival curves are the same for two modes of death (weevil-kill or other-kill) for large class-size T. utriculata ............47 5-5 Median values (months) for failure m odes other deaths and disappearances vs. weevil deaths for large class-size T. utriculata ........................................................47 6-1 Other species of bromeliads mon itored in HHSP, FSSP and SSSP and condition at end of monitoring period......................................................................................60 6-2 Total number of dead bromeliads fallen from canopy into mapped Sections, collected, and categorized as percent killed by M. callizona or killed by cause other than M. callizona .............................................................................................61 A-1 Description and parameters for the five tiers used to define demarcated Sections and Bromeliad Hosts................................................................................................62 B-1 Total hectarage mapped for each Natural Area, mapping and monitoring schedules, and bromeliad species susceptibl e to weevil attack present for each of the five Natural Areas..............................................................................................63 C-1 Size classifications, defined by longest leaf length, for the bromeliad species in this study..................................................................................................................64 D-1 Health rating chart showing the classi fications for the four quarters of the physiological condition scale...................................................................................65
viii LIST OF FIGURES Figure page 1-1 Adult Metamasius callizona .......................................................................................8 1-2 Distribution of Metamasius callizona in south Florida..............................................9 3-1 Weevil count per hectare/month for My akka River State Park, Sarasota County, Florida, from June 2001 to June 2005......................................................................21 3-2 Weevil count per hectare/month for Loxahatchee National Wildlife Refuge (LNWR); Highlands Hammock State Park (HHSP); Fakahatchee Strand Preserve State Park (FSSP); and St. Sebastian Buffer Preserve State Park (SSSP)......................................................................................................................21 4-1 Time series decomposition plot and seas onally adjusted data for â€˜all bromeliadsâ€™ ............................................................................................................................... ...32 4-2 Time series decomposition plot and seas onally adjusted data for large class-size Tillandsia fasciculata ...............................................................................................32 4-3 Time series decomposition plot and seas onally adjusted data for large class-size Tillandsia utriculata .................................................................................................33 4-4 Time series decomposition plot and seasonally adjusted data for average monthly rainfall (cm)...............................................................................................33 4-5 Time series decomposition plot and seasonally adjusted data for average monthly lowest temperature (C)...............................................................................34 5-1 Survival plots for small, medium and large class-size bromeliads..........................44 5-2 Median time to failure with upper an d lower boundaries, according to size and mode of failure.........................................................................................................45 5-3 Survival curves for two failure mode s (death and disappearance, and weevil) for large class-size Tillandsia fasciculata ......................................................................46 5-4 Survival curves for two failure mode s (death and disappearance, and weevil) for large class-size Tillandsia utriculata ........................................................................47
ix 6-1 Status of Tillandsia fasciculata bromeliads in MRSP, LNWR, HHSP, FSSP, and SSSP.........................................................................................................................59 6-2 Status of Tillandsia utriculata bromeliads in MRSP, HHSP, and FSSP.................59 6-3 Status of Tillandsia balbisiana bromeliads in MRSP, LNWR, FSSP, and SSSP....60
x Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science ECOLOGICAL AND DEM OGRAPHIC TRENDS AND PATTERNS OF Metamasius callizona (CHEVROLAT), AN INVASIVE BROMELIAD-EATING WEEVIL, AND FLORIDAâ€™S NATIVE BROMELIADS By Teresa M. Cooper August 2006 Chair: Howard Frank Major Department: Entomology and Nematology In 1989, an immigrant bromeliad-eating weevil from Mexico, Metamasius callizona (Chevrolat), was detected in Broward County, Florida. Despite an eradication attempt, its population increased and spread in the surrounding natural areas. This thesis examines seasonal and ecological trends and patterns of M. callizona and Floridaâ€™s native bromeliads. A multi-tiered method was designed to map areas for monitoring M. callizona and host bromeliads in five Natural Areas in south Florida. Data were collected monthly from June 2001 to June 2005 and includ ed two data sets: 1) Demographic data on a population of selected living bromeliads ; and 2) the collection of fallen dead bromeliads within the monitored areas; the d ead bromeliads were examined for cause of death and for M. callizona specimens. Data collected from data set 1 were used to examine seasonality and survivability in two species of bromeliads, Tillandsia fasciculata Swartz and T. utriculata L (chapters 4 and 5). Data from data set 2 were used to examine seasonal fluctuations in the M. callizona population (chapter 3). Bo th data sets were used
xi to compare habitats and bromeliad communities and to make an assessment on the future of Floridaâ€™s native bromeliads (chapter 6). Conclusions were the following: Metamasius callizona does not exhibit seasonal tr ends, except, perhaps, on T. utriculata . Tillandsia fasciculata exhibited mild seasonal fluctuations while T. utriculata spiked in early spring. Survival analysis showed T. fasciculata to have a higher pro portion of its population killed by M. callizona than by other deaths; T. utriculata had a higher proportion of its population killed by other deat hs than by death caused by M. callizona . Seasonal differences and differences in the response to attack by M. callizona can be explained by the different reproduc tive strategies of T. fasciculata and T. utriculata . Tillandsia fasciculata reproduces vegetatively and by seed, a nd therefore has larger, more stable populations; T. utriculata reproduces only by seed, and therefore has small, more ephemeral populations. Observations on the br omeliad populations from the five Natural Areas showed T. fasciculata to have a higher percentage of survival than T. utriculata ; that T. balbisiana had a low incidence of attack by M. callizona ; and that the population in Fakahatchee Strand Preserve State Park, where no confirmed M. callizona specimens were collected, had an overall high percentage of survival for all species included. The other four Natural Areas ha d an ongoing infestation of M. callizona , and the number of dead, fallen bromeliads collected had a similar percentage killed by M. callizona , 71 â€“ 82%. The future of Floridaâ€™ s native bromeliads may vary depending on the species, and will depend on 1) the increased rate of mortality caused by M. callizona in relation to the speciesâ€™ ability to outgrow or outrun M. callizona ; 2) the range and rar ity of the species; and 3) the distribution of the bromeliad patches.
1 CHAPTER 1 INTRODUCTION In 1989, an immigrant bromeliad-eating weevil from Mexico, Metamasius callizona (Chevrolat), was detected in Broward County, Florida. Despite an eradication attempt, its population increased and spread in the surrounding natu ral areas (Frank and Thomas 1994). Since then, th e weevil has spread across so uth Florida and has been causing great damage to native brome liad populations (Frank and Cave 2005). Metamasius callizona Â’s natural range is in Guatemala and southern Mexico (Frank and Cave 2005). Metamasius callizona is one of 32 known species of bromeliad-eating weevils in the Neotropics (F rank 1999). The adult ranges in size from 11 to 16 mm in length (Frank and Cave 2005) and has a black body with a transverse stripe across its elytra that is colored orange, red, or yellow; rarely, the stripe is not present (Fig. 1). Metamasius callizona is a holometabolous insect a nd, in the laboratory, reared on pineapple stems, the insect had a mean de velopmental time of 57.49 days to grow from within 24 hours of the time the egg was laid to adulthood (Salas and Frank 2001). In the laboratory, after mating once, adult females laid eggs for the duration of their lives, with no periodicity, with an average fecundity of 39.6 eggs per female and an average life span of about 0.5 year (Frank et al. 2006). All life stages of M. callizona live on host bromeliads, wh ich the larvae and adults consume (Frank 1996a). The adults are capable of flying and function as the dispersal unit. Adult weevils consume leaf tissue but do not threaten the life of the host bromeliad; the larva, by mining the meri stematic tissue, kills the pl ant (Larson 2000a). An adult
2 female cuts a slit in the base of a brome liad leaf and inserts an egg. When the larva emerges from the egg, it mines the leaf, but after growing too large, the larva exits the leaf and begins to mine the stem of the plan t (Frank 1996a). This sort of damage results in a characteristic death; the st em of the plant falls out, and the inside is a cavity filled with chewed plant tissue, sometimes cont aining weevil specimens or one to several empty pupal chambers. Florida has one native bromeliad-eating weevil, Metamasius mosieri Barber (Frank and Cave 2005). Not much was known about M. mosieri until after M. callizona entered Florida and began causing damage to the br omeliads, which brought researchers into greater contact with M. mosieri (Frank and Cave 2005). Metamasius mosieri does not cause significant damage to bromeliad populations, as does M. callizona . No specialist predators of M. mosieri have been discovered, but it is speculated that the Florida weevil population is limited in part because it is restri cted to small-sized plants and because it has particular environmental needs (Frank and Cave 2005). Exposur e to sunlight has been shown to be a limiting factor for cer tain species that live in bromeliads. Wyeomyia mitchellii (Theobald), a fly that oviposits in ta nk bromeliads, has been shown to have a preference for bromeliads in shaded habita ts (Frank and OÂ’Meara 1985); and bromeliad communities in general have been shown to be more similar between bromeliads growing under same conditions (sun-exposed or shaded ) than between species of bromeliads (Lopez and Rios 2001). However, the mech anism(s) that limits the growth of M. mosieri populations is yet unknown. It is possible that a specialist natural en emy exists but has not yet been discovered; or th at habitat conditions limit its range (such as sun exposure, humidity levels, and wind exposures).
3 Chemical control of the w eevil is impossible for economic , practical and ecological reasons (Larson 2000b). Because M. callizona lives inside the host bromeliad, in the canopy, huge amounts of chemical would have to be applied to potentially affect the weevil. The amount and dispersal of the chem ical would be outrage ously expensive, and, if it were broadcasted aerially, would harm many non-target species. The only possible method for controlling M. callizona is to use classical biol ogical control. Classical biological control is the pract ice of introducing a natural en emy (usually from the same area that the pest came from) to suppress a pest population. A biological control project was initiated by J. H. Frank and is still ongoing (Frank and Cave 2005). Despite searches in six countries (Mexico, Panama, Honduras, Guatemala, Belize, Paraguay and Peru), only one candidate biological control agent has been found, a new species of tachinid fly, Lixadmontia franki Wood and Cave, from the cloud forests of Honduras, attack ing a related host species , Metamasius quadrilineatus Champion (Cave 1997, Wood and Cave 2006). Tests have shown that the fly will parasitize M. callizona and it is now being reared and tested for suitability for release (Frank and Cave 2005). Test s have also shown that L. franki will parasitize M. mosieri , but that it has a preference for M. callizona over M. mosieri ; probably because M. callizona is larger than M. mosieri and closer to the size of M. quadrilineatus . Bromeliads (Bromeliaceae) originated in the Neotropics, estimated about 60 million years ago near the end of the Cr etaceous period (Benzing 1980; Isley, 1987). Most native bromeliad lands are in South and Central America; only a few species are found in the southeast porti on of North America (Benzing 1980; Isley 1987). Three genera ( Tillandsia , Catopsis and Guzmania ) in the subfamily Tillandsioideae are found in
4 Florida (Isley 1987). All of FloridaÂ’s nativ e bromeliads (16 species in total) are epiphytes and grow in hardwood, pine, and swamp forests (Benzing 1987; Larson 2000c). Twelve of the 16 species of br omeliads are susceptible to attack by M. callizona (Frank and Cave 2005). Most bromeliads are frostintolerant and, therefore, the number and total biomass of bromeliad species drops dramatically from subtropical south Florida to north Florida, where seasonal temperatures are more pronounced and freezes are greater in number and duration (Benzing 1980; Myers and Ewel 1990). FloridaÂ’s bromeliads require sufficiently humid conditions in order to su rvive and are more abundantly found over or near water, along rivers or stream s or in swamps (Larson 2000c). The spread of M. callizona is dependent on the range of its host plants. Since 1989, M. callizona has spread across most of south Florida (Fig. 2; Ferriter 2006) and it is likely that the weevil will continue to spread until it has covered the range of all potential host bromeliads (Frank and Cave 2005) . While, since its arrival, M. callizona has been causing great damage to wider ranging species of bromeliads, particularly Tillandsia fasciculata Swartz and T. utriculata L., it has recently been found in the Big Cypress National Preserve (Frank 1996b). The Big Cypr ess National Preserve and the Everglades in general support the greatest diversity of bromeliads in Florida, including the wider ranging species as well as several rare species ( T. flexuosa Swartz, T. paucifolia Baker, T. pruinosa Swartz, Guzmania monostachia L., Catopsis berterioniana Schultes, C. floribunda Brongniart, and C. nutans Swartz; Larson 2000c). The arrival of M. callizona into this area heralds the potential for the loss of numerous species of bromeliads (Frank and Cave 2005).
5 Conservation of FloridaÂ’s bromeliads is not a trivial matter. In a worse case scenario, 12 species of bromeliads native to Florida could be lost, along with the ecosystems of vertebrate and invertebrate animals these plants support (Frank and Cave 2005). All of FloridaÂ’s native bromeliads associ ate with other organism s to create small, complex ecosystems. Some species, such as T. pruinosa and T. balbisiana , have mutualistic associations with ants (Benzi ng 1980). Many tank bromeliads, such as T. utriculata , T. fasciculata , and G. monostachia , hold water in the axils of their leaves, which support aquatic ecosystems (Frank 1983). Numerous arthropods have been collected from tank bromeliads but many have yet to be identified and very little is known about the dynamics of these associations (Frank et al. 2004). Bromeliads also support canopy animals such as raccoons, snakes, rodents, and birds (Butler 1974). Bromeliads are a source of water during the dr y Florida winters. They are a food source for phytophagous insects (other than M. callizona , and which are usually not harmful to the plant in the l ong run). These phytophagous insects serve as prey for carnivorous animals. Bromeliads offer a place for many animals to build a home, or to provide a base upon which to build a nest (Frank 1983, Butler 1974). Bromeliads are pivotal to lif e in the canopy, and as such th ey are an important part of nutrient cycles (Benzing 1980). Bromeliads are adapted to abso rb nutrients through their leaves. They obtain their nutrition fr om rainwater that has leached through the canopy, from the waste material of the organism s that live in and on them, from insects or other small animals that die in the plant, from detritus that falls from the canopy, and from algae that grow in the tank water (Frank 1983).
6 Besides these very important ecological roles, bromeliads offer excellent opportunities for ecological experiments, such as studying nutrient cycles (Nadkarni 1992, McNeely 1998). Other studies that woul d be useful include Â“island studiesÂ” (bromeliads are similar to islands since many of the species they support must find ways to migrate from one bromeliad to another bromeliad, across generations) and ecological succession (bromeliads are much more slow growing than the ecosystems they support, and as the plant grows from seedling to adul t to death, the ecosystem changes). The ecosystems supported by bromeliads provide a format for studying trophic levels, as well as numerous species, many yet to be identifi ed, for studying systematics. Bromeliaceae are a large, diverse family whose member s are good candidates fo r studying evolution. Bromeliads can be used to monitor pollution (Benzing 1991). As a bromeliad absorbs nutrients through its l eaves, it will also absorb air borne pollutants. Bromeliads would make better indicators than lichens becau se bromeliads are vascular plants, similar to most of the natural vegetation in Flor ida (Benzing 1991). Urbanization and energy demands are going to continue to grow in s outh Florida; bromeliads could be used to monitor the effect such growth is having on surrounding natural areas . Bromeliads can also be used to monitor climate change (Na dkarni 1992). If temperat ures rise, their range should expand; and, conversely, if temper atures drop, their ra nge should contract. Bromeliads serve as an excellent avenue for educating the general public about conservation efforts, as well as getting the pub lic involved in the process. Bromeliads are an asset to FloridaÂ’s State Parks (Larson 2000d). Many peopl e visit FloridaÂ’s parks to photograph or to create paintings of the brome liads, or just to drive or walk among them and find pleasure in their presence.
7 In an effort to help conserve FloridaÂ’ s native bromeliads and to assist in the biological control project that was initiated to control M. callizona , this thesis looks at seasonal trends and patterns for M. callizona and some of the native bromeliads that act as its host. As well, survival analysis was used to determine the effect that M. callizona is having on two bromeliad species, T. fasciculata and T. utriculata . Research was performed in five Natural Areas: Myakka River State Park, Loxahatchee National Wildlife Refuge, Highlands Hammock State Park, Fakahatchee Strand Preserve State Park, and St. Sebastian Bu ffer Preserve State Park.
8 Figure 1-1: Adult Metamasius callizona . Photo: J. C. Yawn.
9 Figure 1-2: Di stribution of Metamasius callizona in south Florida (Ferriter 2006).
10 CHAPTER 2 METHOD AND MATERIALS This method was designed to monitor bromeliad populations in south Florida and the invasive bromeliad-eating weevil, Metamasius callizona (Chevrolat) that arrived in Florida in 1989 and that has since been attacking FloridaÂ’s native bromeliads. Bromeliads were monitored in five Natura l Areas in south Florida: Myakka River State Park (MRSP), Loxahatchee Nationa l Wildlife Refuge (LNWR), Highlands Hammock State Park (HHSP), Fakahatchee St rand Preserve State Park (FSSP), and St. Sebastian Buffer Preserve State Park (SSSP) beginning in June 2001 and ending in June 2005. Myakka River State Park was chosen as the primary research site because it had a large bromeliad community which included both Tillandsia fasciculata Swartz and T. utriculata L. (two of the large bromeliad species that have a wide range in south Florida and that have been under heavy attack by M. callizona ) and because M. callizona had only recently been found in the park (in Sept ember 2000, one year before the start of this research; Frank 1996b). The other four Natu ral Areas were selected in order to incorporate localities with different species of bromeliads into th e study and based on the availability of volunteers to collect data. Data were collected from demarcated S ections that were defined using a multitiered method. Tiers were defined by brome liad habitat and Bromeliad Host density. Bromeliad Hosts are the substrates upon which the bromeliads grow; usually a host was a tree, however some were vines or stumps, and a few of the bromeliads grew terrestrially.
11 The tiers, by decreasing size, were calle d Natural Area, Region, Area, Section, and Bromeliad Host (see Table A-1 in Appendix A). Sections were mapped using a baseline from which to triangulate trees, Bromeliad Hosts, and other landmarks. These number s were transcribed to grid-mapping paper using a compass and straight-edge. The maps we re used to define the boundaries for data collection and to calcula te the area monitored. Materials for mapping a Section included tw o 91 m surveyor tapes; one 1.5 m stick with a small portion sanded level on the b ack end, with two pegs stuck in to which a compass could be affixed; a compass with holes drilled in it to fit the pegs on the stick; surveyor flags to mark Bromeliad Hosts; stak es to hold down surveyor tapes; and a log book for recording data. To map a Section, a baseline was laid along so me definable path (an actual path or road, or a chosen stretch in a stand of trees). Compass bearings were taken for the baseline. The objects to be mapped (the tr ees, including the Brom eliad Hosts and other obvious, persistent landmarks) were positioned by triangulating each object, using at least two compass bearings from the baseline. R ough estimations show this method of mapping is accurate to within approximately a third of a meter; the further the objects were from the baseline, the greater the loss in accur acy. Information recorded for each object included compass bearing and baseline re ading for each line of sighting, the type (hardwood, pine or palm) and size of the tr ee (or other object); and whether it was a Bromeliad Host.
12 Two data sets were collected monthly within the demarcated Sections: 1) Demographic data on a population of selected living bromeliads growing in the canopy; and 2) collection of fallen dead brome liads within the demarcated Sections. For data set 1, a portion of the Bromeliad Hosts within each Section was randomly selected and the bromeliads on these Br omeliad Hosts made up the population of bromeliads being monitored. Each Bromeliad Host was sketched from a particular spot (indicated on the Section map), and its resident bromeliads were added in the sketch for relocation purposes. Each bromeliad was a ssigned a unique number and identified to species (if possible; T. utriculata and T. fasciculata are difficult to distinguish when they are small and medium-size plants). Data coll ection included class-size (small, medium or large). Class sizes were based on longest leaf length. Botanists ha ve traditionally used leaf area and dry weight to measure size of plants. The di sadvantage of these measurements is that they requi re killing the plant. Length of longest leaf was developed as a non-lethal method for measuring T. utriculata and was related by regression to water-impounding capacity of leaf axils (Frank and Curtis 1981). Le ngth of longest leaf was measured together with dry weight a nd leaf area by Frank et al. (2004). Size classifications vary according to specie s; Table C-1 in Appendix C gives the size classifications used in th is study for species included. Health ratings were assigned monthly for the selected bromeliads. The health rating was an indicator of the condition of the bromeliadâ€™s health and was based on assessment of certain outwardly physical charact eristics, such as the color and fullness of the leaves and physical injuries that were visible. Health ratings ranged from 3.0 (thriving and well; no injuries, discolorati on, or dehydrated leaves) to 1.0 (completely
13 dead; no green left remaining on the plant). The 3.0 to 1.0 range was divided into four quarters: 3.0 to 2.5 (healthy); 2.4 to 2.0 (moderately stressed); 1.9 to 1.5 (heavily stressed); and 1.4 to 1.0 (seriously stressed; ultimately ending in death). Table D-1 (Appendix D) outlines these four quarters and the characteristics that indicate where a bromeliad would be categorized. Bromeliads were chosen based on appare ncy; if a Bromeliad Host was covered with bromeliads, this was noted, but only the most obvious were mapped. When a selected bromeliad died or disappeared, evid ence would be sought to determine the cause of death or disappearance. To replace dead or lost bromeliads, Bromeliad Hosts were updated and replaced every six months. If all of the bromeliads being monitored on a Bromeliad Host died or disappeared, then the Bromeliad Host was examined thoroughly for any remaining bromeliads, and these we re monitored; if there were no more bromeliads available, then a new Bromeliad Ho st in the Section was randomly selected to take its place. For data set 2, the Section maps defined the area on the ground to be searched for dead bromeliads that had fallen from the canopy. The bromeliads were examined for cause of death. If weevil specimens or pupal ch ambers were present, they were collected. An attempt would be made to rear larvae a nd pupae to adulthood because it is difficult to distinguish M. callizona larvae and pupae from that of Fl oridaâ€™s native bromeliad-eating weevil, M. mosieri Barber. MRSP, the primary research site, was ma pped first; subsequently, five new Areas were added to existing Regions, and one new Region was added. The other four Natural Areas were included later than June 2001 a nd ended before June 2005. Only LNWR had
14 a new Area added after its initial mapping. MR SP data were collected by me. The other Natural Areas were mapped by me, but monitored by volunteers using my method. Table B-1, in Appendix B, shows the total hectarage mapped for each Natural Area, the mapping and monitoring schedules, and the bromeliad species that were monitored. The five Natural Areas that were monitored represented different habitats and bromeliad communities. MRSP was monitored monthly for 49 months. Total land area monitored in MRSP covered 2.98 hectares. Seventeen Ar eas were mapped in hardwood forests, hammocks and mixed hardwood/palm forests. Seven hundred thirty-nine bromeliads were selected for monitoring; the populat ion was composed of approximately 72% Tillandsia fasciculata Swartz, 27% T. utriculata L., and 1% T. balbisiana Schultes. LNWR was monitored for 28 months until monitoring was interrupted by hurricane activity in August and September 2004. Total land area monitored was 0.05 hectares and included 115 selected bromeliads. Three ar eas were mapped, one in the interior of a cypress dome and two on the edge of the c ypress dome. The bromeliad population was approximately 81% T. fasciculata and 19% T. balbisiana . At the start of the study, no T. utriculata bromeliads were apparent in the c ypress dome. Wildlife Biologist Marian Bailey informed me that T. utriculata used to be present in th e Refuge, but she only knew of one remaining specimen, located near th e Administrative Building; it was infested with M. callizona and soon died as a resu lt of the infestation. HHSP was monitored for 33 months. Total land area monitored was 0.45 hectares in hardwood forest; two Areas were mappe d, one in an orange grove and one in hardwood forest. Twenty-one bromeliads we re selected for monitoring, of which 33% were T. fasciculata , 13% were Tillandsia simulata Small, and 54% were T. utriculata .
15 Tillandsia simulata is precinctive to Florida (Larson 2000c). Only two T. simulata plants were spotted in the monitoring area, making this a small sample. Tillandsia variabilis Schlechtendal was present in HHSP but not in the Sections that were monitored. FSSP was monitored for 24 months. Tota l land area monitored was 0.04 hectares and included four Areas, three in swamp fore st, and one on the side of a service road running through swamp forest. Se venty-seven bromeliads were selected for monitoring; the population was composed of 32% T. fasciculata , 10% T. utriculata , 12% T. balbisiana , 29% T. pruinosa Swartz, 3% T. variabilis , and 14% Guzmania monostachia L. The three Areas in the swamp were sm all patches each with a dominant species ( T. utriculata , T. pruinosa , and G. monostachia ); the Area on the service road was a combination of the species listed for FSSP. The percent composition of the species given here is likely not representa tive of the surrounding land because FSSP has more species than those included in this study ( T. flexuosa Swartz, T. paucifolia Baker, Catopsis berterioniana Schultes, C. floribunda Brongniart, and C. nutans Swartz; Larson 2000c), and because the bromeliads tend to grow in pa tches such as the three out of four Areas mapped in FSSP. SSSP covered 0.57 hectares; three Areas were mapped, one each in a cypress dome, a swamp forest, and an oak hammock. One hundr ed eighteen bromeliads were selected for monitoring and included 78% T. fasciculata , 10% T. balbisiana , 11% T. paucifolia , and 1% T. simulata . Tillandsia paucifolia were located on the edge of the cypress dome and co-existed with small and medium sized T. fasciculata and T. balbisiana .
16 CHAPTER 3 DOES Metamasius callizona (CHEVROLAT) EXPERIENCE SEASONAL POPULATION FLUCTUATIONS? Introduction In 1989, an immigrant bromeliad-eating weevil from Mexico, Metamasius callizona (Chevrolat), was detected in Broward County, Florida. Despite an eradication attempt, its population increased and spread in the surrounding natu ral areas (Frank and Thomas 1994). Since then, th e weevil has spread across so uth Florida and has been causing great damage to native brome liad populations (Frank and Cave 2005). Previous surveys for M. callizona had been undertaken to demarcate the everincreasing area occupied by it, and to colle ct specimens for laboratory use (Frank and Thomas 1994, Frank 1996b). Seldom was a site revisited once weevils were found, so the data could not readily reveal se asonal trends in abundance. This chapter, in contrast, is a result of repeated visits to a select number of sites. It evaluates evidence for seasonal trends of M. callizona populations in five na tural, protected areas. Method and Materials Metamasius callizona specimens were collected in Myakka River State Park (MRSP), Loxahatchee National Wildlife Re fuge (LNWR), Highlands Hammock State Park (HHSP), Fakahatchee Strand Preserve State Park (FSSP), and St. Sebastian Buffer Preserve State Park (SSSP). MRSP was the primary research site and monitoring there began in June 2001 and ended in June 2005. Monitoring began in the other four Natural Areas after June 2001 and ended before June 2005. Table B-1 in Appendix B outlines the
17 mapping schedule for the five Natural Areas, as well as the hectarage mapped and the bromeliad species included in the study. Weevil larvae, pupae and adults were coll ected from dead bromeliads that had fallen from the forest canopy within demarcated Sections. These demarcated Sections were defined using a multi-tiered method (see chapter 2 for details on this method and Table A-1 in Appendix A that defines the Tiers). Monitoring was conducted monthly. All d ead bromeliads susceptible to weevil attack that fell into a selected Section were opened and examined. The number of weevil specimens found in a Section was recorded al ong with the date, time, life stage of the specimen (adult, pupa or larva; no eggs were collected), and whethe r the weevil specimen was alive or dead. If possible, larvae a nd pupae were reared to adulthood for species identification. The average number of living adults a nd living or dead pup ae and larvae found for each Natural Area was divided by the hectarage monitored for each Natural Area and plotted against time. The weevil population wa s treated as a Poisson distribution because it was not dense and was not randomly distri buted. Upper and lower boundaries (plus and minus 2 standard errors) for each plot we re included. The plots were examined for peaks that regularly rose above the uppe r boundaries of a baseline population size, indicating seasonal fluctuation. Results and Discussion None of the plots indicate seasonal fluctuations in the weevil population. Metamasius callizona was found in all seasons and no p eaks rose significantly higher than any other (exceeding the upper boundaries). Fig. 3-1 shows the average number of M. callizona specimens found per hectare plotted against time (months) for Myakka
18 River State Park (MRSP). Fig. 3-2 show s the average number of weevils found per hectare plotted against time (months) for the other four Natural Areas. These results indicate that, overall, the weevil population remains constant and is active throughout the year. With an estimat ed mean generation time of 13 Â– 17 weeks (Salas and Frank 2001; Frank and Thomas 1994), M. callizona could potentially have three to four generations per ye ar. In the laboratory, after mating once, adult females laid eggs for the duration of their lives, with no pe riodicity, with an average fecundity of 39.6 eggs per female and an averag e life span of about 0.5 year (F rank et al. 2006). The lack of periodicity supports the conclusion herein that the weevils do not experience seasonal fluctuations. Constant activity, multiple generations per year, single mating with high fecundity, and long life span wit hout regulation have contribute d to the fast expansion of M. callizona populations, and to the high levels of destruction incurred upon native bromeliad populations by M. callizona . Lack of seasonality is probably due to the subtropical climate that exists in south Florida, and to the fact that the weevil lives inside its hos t plant, where it is protected from extreme changes in the weather. Ther e are wet and dry seas ons in south Florida (Myers and Ewel 1990), however, most of Fl oridaÂ’s bromeliads undergo only moderate fluctuations in physiological condition (flowering excepted) with these changing seasons (see chapter 4). Tillandsia utriculata , one of FloridaÂ’s bromeliads that are susceptible to weevil attack, has suffered greatly since 1989. Unlike FloridaÂ’s other native bromeliads, T. utriculata does not propagate vegetatively, but di es after producing seeds (Benzing 1980; Isley 1987); and tends to go th rough growth spurts with the onset of spring (see chapter
19 4). In MRSP, of the 13 large, weevil-killed T. utriculata found containing weevil specimens, 10 fell at the beginning of the rai ny season (May, June, and July) and had an average of 6.2 weevil specimens per brome liad, including living and dead larvae and pupae, and living adults. Three infested plants were found outside the rainy season (February, October and March, one per m onth) and each contained only one weevil specimen. The sample of large T. utriculata bromeliads in MRSP was small, and therefore does not offer conclusive evidence of weevil seasonality on this particular host plant of this particular class-size. However, with the observed growth spurts in the spring and the higher average number of weevils found during the months of May through July, it is possible that in a large patch of infested T. utriculata plants the weevil may exhibit seasonality. A study that examines a large T. utriculata population (such as a wellestablished colony with at le ast a few hundred bromeliads and at the forefront of an infestation) would be necessary to determine the accuracy of this statement. The other Natural Areas studied did not have large T. utriculata populations. A large amount of land must be covered in order to collect a me aningful number of weevil specimens. Pupal chambers in an infested area can be found in much larger numbers than adults, larvae or pupae; however , the pupal chambers are highly persistent and cannot be used to determine seasonalit y. Weevil specimens were collected from LNWR, HHSP, and SSSP, but in very low numb ers (total numbers collected for each site were 4, 5, and 11, respectively). The greate st number of weevil specimens (132) was found in MRSP, where more bromeliads were surveyed over a longer time than at the other Natural Areas (Table B-1) . Weevil specimens were collected regularly for the first
20 two years (from June 2001 to June 2003, a to tal of 113 weevil speci mens was collected) but declined in the final two years (from Ju ly 2003 to June 2005, only 19 were collected). This may be due to the loss of bromeliads, especially large T. utriculata , in the Sections that were being monitored. Fluctuations in the weevil population are not seasonally affected (except, perhaps, in the case of T. utriculata ). Weevil populations are more likely to vary based on the number and species of bromeliads that make up a patch, the size and density of the patch, the ability of the patch to support the weevil population, and whether an infestation is newly initiated or has been present for some time.
21 Figure 3-1: Weevil count per hectare/mont h for Myakka River State Park, Sarasota County, Florida, from June 2001 to June 2005. Figure 3-2: Weevil count per hectare/mont h for Loxahatchee National Wildlife Refuge (LNWR); Highlands Hammock State Park (HHSP); Fakahatchee Strand Preserve State Park (FSSP); and St. Sebastian Buffer Preserve State Park (SSSP).
22 CHAPTER 4 SEASONAL TRENDS AND PATTERNS OF Tillandsia fasciculata SWARTZ AND Tillandsia utriculata L. POPULATIONS IN MY AKKA RIVER STATE PARK Introduction Florida has 16 native species of br omeliads, including three genera ( Tillandsia , Catopsis and Guzmania ) in the subfamily Tillandsioideae (Isley 1987). All of FloridaÂ’s native bromeliads are epiphytes and grow in various hard wood, pine, and swamp forests (Benzing 1980). The canopy is a harsh enviro nment, subjected to drought conditions, low nutrition availability, and catastrophic even ts. Bromeliads have adapted to these conditions through vegetative reduction, nove l growth habits, and the evolution of specialized trichome cells that absorb and secure water (Benzing 1980; Isley 1987). Most bromeliads are frost-in tolerant and, therefore, temp eratures limit the range of most bromeliads (Benzing 1980). In Florida, the number and total biomass of bromeliad species drops dramatically from subtropica l south Florida to north Florida, where seasonal temperatures are more pronounced and freezes are greater in number and duration (Benzing 1980; Myers and Ewel 1990). FloridaÂ’s bromeliads require sufficiently humid conditions in order to su rvive and are more abundantly found over or near water, along rivers or stream s or in swamps (Larson 2000c). Rainfall regulates timing for emergence of the inflorescence and the release of seed (Benzing 1980). FloridaÂ’s bromeliads pr oduce inflorescences between midwinter and spring, seed develops over th e span of a year and, in the following spring, the seeds dehisce and are released just before the spring rains begin (Benzing 1980). This timing
23 ensures that the wind-dispersed seeds are dry at the time of dispersal, and that once the rains follow shortly thereafter, they are attached to a potential host. Tillandsia paucifolia Baker provides an example of how important the timing of seed release relates to rainfall. On Sanibel Island off the southwest coast of Florida, the rainy season starts 4-6 weeks after the season begins on the mainland; T. paucifolia populations on the island release seed 4-6 weeks later than mainland T. paucifolia populations, in time with the rainfall (Benzing 1980). Twelve of Floridaâ€™s native bromeliads ar e susceptible to attack by an invasive bromeliad-eating weevil Metamasius callizona (Chevrolat). Field research has shown that M. callizona exhibits no seasonal population fluctu ations, with the possible exception on the host bromeliad Tillandsia utriculata L. (see chapter 3). This chapter examines the bromeliads that are susceptible to attack by M. callizona in Myakka River State Park (MRSP) for seasonal patterns and trends in re lation to seasonal changes in rainfall and temperature. The two primary species of bromeliads in this study were Tillandsia fasciculata Swartz and T. utriculata . Both species are large, long-lived plants (Isley 1987) with a wide distribution ac ross the southern peninsula. Tillandsia balbisiana Schultes was also present in the study, but was a very small portion of the bromeliad community in MRSP; only two were include d in the selection of bromeliads for monitoring. Methods and Materials From June 2001 to June 2005, 739 bromeliads were monitored monthly in Myakka River State Park (MRSP). Bromeliads in th e study included those species susceptible to attack from M. callizona , which were T. fasciculata , T. utriculata , and T. balbisiana , and included class-sizes ranging from very small to very large.
24 Size classifications, based on longest leaf length, for T. fasciculata and T. utriculata were: Small, 15 cm; medium, 15 to 60 cm; and large, > 60 cm. Size classifications for T. balbisiana were: Small, 5 cm; medium, 5 to 15 cm; and large, > 15 cm; the two plants included in this study we re large. Defining characteristics between T. fasciculata and T. utriculata are difficult to distinguish when they are small and medium class-size; as they become large, they are very easy to tell apart. Therefore, only large class-size bromeliads we re used in this analysis. The bromeliads selected for monitoring in MRSP were chosen based on a multitiered method. (See chapter 2 for details on this method and Table A-1 in Appendix A that defines the Tiers.) Monitoring was conducted monthly and consis ted of assigning a Â“health ratingÂ” to the individual bromeliads in the study. The health rating was an indicator of the condition of the bromeliadÂ’s health and wa s based on assessment of certain outwardly physical characteristics, such as the color and fullness of the leaves and physical injuries that were visible. Health ratings ranged from 3.0 (thriving and well; no injuries, discoloration, or dehydrated l eaves) to 1.0 (completely dead ; no green left remaining on the plant). The 3.0 to 1.0 range was divided into four quarter s: 3.0 to 2.5 (healthy); 2.4 to 2.0 (moderately stressed); 1.9 to 1.5 (heavily stressed) ; and 1.4 to 1.0 (seriously stressed; ultimately ending in death). Ta ble D-1 in Appendix D outlines these four quarters and the characteristics that indicat e where a bromeliad would be categorized. A health rating was assigned to each bromeliad, along with an explanation for the health rating (e.g., core falling out of the center of the plant, heavy discoloration, etc.) in
25 order to determine cause of stress or injury. If a bromeliad died or disappeared, evidence was sought to determine the cause of death or disappearance. Decomposition analysis, us ing MinitabÂ® Release 14.20 (Minitab, Inc. 2006), was performed on the following data sets: All monitored bromeliads ( Tillandsia fasciculata , T. utriculata and T. balbisiana , all sizes; n = 739); large cl ass-size (longest leaf length > 60 cm; n = 109) of T. fasciculata ; large class-size (longest leaf length > 60 cm; n = 41) of T. utriculata ; the average monthly rainfall; and average monthly lowest temperature (rainfall and temperature data were collected in the Bradenton/Sarasota area by the National Weather Service; National Weather Se rvice 2005). Data for each set (the three bromeliad sets; rainfall data and lowest te mperature data) were decomposed into trend and seasonal components using a multiplicative model to get median raw seasonal values for the twelve months of each year. Raw seas onal values were used to calculate seasonal indices for adjusting the data (setting the median of the raw seasonal values to equal 1) and to plot boxplots for comparing the medi ans and variance for each month. Trend lines for each data set were calculated using least squares regression. Cross correlations were made for the thr ee bromeliad data sets (Â‘all bromeliadsÂ’; large class-size T. fasciculata ; and large class-size T. utriculata ) with the average monthly rainfall and with the average lowest te mperature. The number of lags was set at +/-17 using the calculation (sqrt (n)) +10; n=length of the tim e series (49 months). Alpha level (5%) was calculated with the formula = 2/sqrt(n-[k]) where n = length of the time series and [k] = the absolute value for each lag time. The cross correlation coefficient (CCF) was determ ined for each lag time. The CCF with the
26 greatest significant difference was tabulate d with its corresponding lag time and alpha value. Results and Discussion Â‘All bromeliadsÂ’, large T. fasciculata , and large T. utriculata all had increasing trends over the four year study (see Fig. 4-1a,42a, and 4-3a). All bromeliads and large T. fasciculata had moderately rising trends with an increase of about 0.1 in the health rating. Large T. utriculata had a more pronounced rise in trend, increasing its average health rating by 0.3 points. Rainfall had a slightly rising tre nd with the average increasing by 5 cm. Average lowest temperatur e had a slightly decreasing trend with an average decrease of 2Â°C. Seasonal rainfall does not have a smooth declin e and rise; rather, average rainfall is significantly greater (and exhibi ts higher variability) in Ju ne, July, August and September compared to the months from October to Ma y (Fig. 4-4b). In October, average rainfall drops suddenly, with little variation in this m onth. Most of the wint er and early spring months show little variability, except for D ecember and, to a lesser extent, November. Â‘All bromeliadsÂ’ and large T. fasciculata show improvements in physiological condition with increased rainfall in June. Seasonal average lowest temperature begins decreasing in October, reaches a low point in January, and then begins to rise. Overall, there is little variation; variation is greatest in December and January. The seasona l plot for lowest average temperature is very similar to the seasonal plot for the heal th ratings of Â‘all bromeliadsÂ’ (Fig. 1b) and large T. fasciculata (Fig. 2b). Health ratings for large T. utriculata , which have similar health ratings from June to February, in creases significantly in March, April and May
27 (Fig. 4-3b), when the average lowest temperat ures are rising; howev er, it does not follow the increase in temperature closely. Highest cross correlation factors (CCF s) for â€˜all bromeliadsâ€™, large T. fasciculata and large T. utriculata at their respective lag times ar e higher than their associated P values (Table 4-1); therefore, the null hypot hesis that the cross correlation for the lag time is equal to zero is rejected (i.e. th ere is correlation). â€˜All bromeliadsâ€™, T. fasciculata and T. utriculata all have higher CCFs for temperature than for rain. All bromeliads and T. fasciculata have much more significantly higher CCFs than T. utriculata , whose CCFs are only slightly larger than the associated P -values. Lag times for â€˜all bromeliadsâ€™ and T. fasciculata are -1, except for â€˜all bromeliadsâ€™ versus average rain, which is zero. These num bers are consistent wi th the graphs, where often the response of the bromeliads lags a month behind, or responds in time with, the changes in temperature and rainfall. In Fig. 4-1b (â€˜all bromeliadsâ€™) and 4-2b (T. fasciculata), the median health ratings are simila r from September to October while temperature falls. As temperature continue s to decline (November and December), â€˜all bromeliadsâ€™ starts to decline as well, while T. fasciculata continues to maintain similar health ratings and does not begi n to decline until January in to February and March, when temperatures start to rise. The health ratings for â€˜all bromeliadsâ€™, as well, decline from January to March. The rest of the year, health ratings for â€˜all bromeliadsâ€™ and T. fasciculata health ratings follow the temperature as it rises. In Fig. 4-1b and 4-2b, â€˜all bromeliadsâ€™ and T. fasciculata maintain similar health ratings from September to November when the average rainfall suddenly drops. â€˜All bromeliadsâ€™ follows rainfall pattern more clos ely throughout the rest of year. This is
28 probably because Â‘all bromeliadsÂ’ includes sm all and medium-sized bromeliads as well as large bromeliads. Small and medium brom eliads respond more readily to changes in environmental extremes due to their smaller mass and their inability to hold as much water or nutritional debris (Benzing 1980). Health ratings for the large T. fasciculata increase in the spring before an increase in rainfall in April and May. The health ratings for T. utriculata have a sharp increase in health ratings in March, April and May, before the increase in rainfall. The ability of the large bromeliads to show improved physiological condition before an increase in rainfall may be a reflection of their le sser dependence on rainfall be cause they can hold water in their tanks for many months (Benzing 1980) and/or nutrition acquisition since the tank water holds animals and plants, both living a nd dead, that contribute to the nutrition of the bromeliad (Frank 1983). The large br omeliads may be affected more by the decreased winter temperatures than by lowe r rainfall, but once temperatures rise, new growth occurs and damage d cells can be repaired. However, T. fasciculata Â’s health ratings continue to rise into the summer as the temperatures rise. The health ratings for T. utriculata spike in the springtime, then fall to a lower level where they re main throughout the year, seem ingly not too affected by rainfall or temperature (note the CCFs in Tabl e 4-1 which are not as significantly high as those for Â‘all bromeliadsÂ’ or T. fasciculata ). Also, the lag times for T. utriculata are -8 (rainfall) and 3 (temperature), whic h would indicate a response from T. utriculata (an increase in health rating) eight months after the summer rains, or three months before the increase in temperatures, ne ither of which is sensible.
29 Both T. fasciculata and T. utriculata are monocarpic and die after going to seed; however, T. fasciculata also propagates vegetatively, and grows in large clumps. Tillandsia utriculata reproduces primarily by seed and only very rarely produces offsets (Isley 1987). These two life strategies result in different life hi stories (Benzing 1980). Because T. fasciculata partitions its resources, an individual plant releases less seed which results in fewer new colonies; however , the offsets retain a secure perch and, because an offset grows rapidly on the nutrition of its dying parent, it reaches reproductive age much more quickly (one to two years compared to 10 to 15 years from seedling). Tillandsia utriculata puts all of its resources into making seed. Of the 41 large T. utriculata in this study, 23 produced an inflorescen ce; 83% of the time, the first sign of inflorescence growth happened in March, Apri l and May, the same months in which the health ratings spiked. (The other 17% ha ppened in January, Febr uary, June, and July, with only 1 spike per month.) Increase in temperature (or some other environmental variable, such as day length) may act as an environmental cue to the plant to maximize nutritional uptake for the development of s eed, which will occur over the following year (seeds were released in March and April). Interestingly, of the 13 large T. utriculata bromeliads in MRSP that were killed by M. callizona and that contained weevil specimens ( living adults, living or dead pupae or larvae), the highest weevil specimen to br omeliad ratio found from this dead population occurred in May (four dead bromeliads w ith 7.5 weevil specimens per bromeliad), June (five dead bromeliads with 6.6 weevil spec imens per bromeliad), and July (one dead bromeliad with three weevil specimens). Th e remaining three dead bromeliads were
30 found in February, March, and October, re spectively, each with a weevil specimen to bromeliad ratio of one. When reared on pineapple tops in the laboratory, M. callizona required an average of 57.5 days to develop from egg to adult (Salas and Frank 2001). Allowing for longer developmental time in the field, an egg laid in March, when T. utriculata â€™s health ratings spiked, could reach adulthood within 8 to 16 weeks (May to July), the same months with the highes t number of weevil specimens to bromeliad. Metamasius callizona showed no seasonality (see chap ter 3), except in this small population of large, weevil-killed T. utriculata bromeliads. It may be that M. callizona is exploiting increased nutritional acquisition that the plant would have used for growth and/or seed production. Howe ver, the number of dead T. utriculata containing weevil specimens was relatively small (n = 13). To determine whether M. callizona exhibits seasonality on this class-size of this species would require further st udies, preferably on a large, well-established T. utriculata colony with at least a few hundred bromeliads and at the forefront of an infestation. Fluctuations in the health ratings for â€˜all bromeliadsâ€™, large T. fasciculata , and large T. utriculata remained, for the most part, in the u pper quarter (3.0 to 2.5) and just dipped into the 2nd quarter (2.4 to 2.0) during the winter mont hs (see Figs. 1b, 2b, and 3b). Bromeliads tend to die quickly, or to not s how outward symptoms of internal damage (such as by M. callizona larval chewing) until the plant is at the point of falling out of the canopy. The exception to this is when T. utriculata dies from going to seed; that is a long process that happens over the span of th e year when the seed is developing. In conclusion, the bromeliad population as a whole (all bromeliads) and the large class-size of T. fasciculata bromeliads showed definite seasonal patterns and had
31 significantly high CCFs with average monthly rainfall and with average lowest temperatures. For both data sets, temper ature had higher CCFs and followed more closely the seasonal patterns of th e bromeliads than did rainfall. Tillandsia utriculata spiked seasonally, in the spring months, a nd had significantly high CCFs, but not much higher than the associated alphas; the lag times for T. utriculata were nonsensical. The health of large class-size T. utriculata is not affected much by changes in temperature or rainfall.
32 Figure 4-1: Time series decomposition plot and seasonally adjusted data for Â‘all bromeliadsÂ’. a) Time series deco mposition plot (n=739). b) Seasonally adjusted data using seasonal indices to plot medians and variation for each month. Figure 4-2: Time series deco mposition plot and seasonally ad justed data for large classsize Tillandsia fasciculata . a) Time series decomposition plot (longest leaf length > 60 cm; n=110). b) Seasonally adju sted data using seasonal indices to plot medians and variation for each month.
33 Figure 4-3: Time series deco mposition plot and seasonally ad justed data for large classsize Tillandsia utriculata . a) Time series decomp osition (longest leaf length > 60 cm; n=41). b) Seasonally adjusted data using seasonal indices to plot medians and variation for each month. Figure 4-4: Time series d ecomposition plot and seasonally adjusted data for average monthly rainfall (cm). a) Time series decomposition plot; data collected by the National Weather Service in the Brad enton/Sarasota area in south Florida (National Weather Service, 2005). b) S easonally adjusted data using seasonal indices to plot medians and variation for each month.
34 Figure 4-5: Time series d ecomposition plot and seasonally adjusted data for average monthly lowest temperature (C). a) Time series decomposition plot; data collected by the National Weather Servi ce in the Bradenton/Sarasota area in south Florida (National W eather Service, 2005). b) Seasonally adjusted data using seasonal indices to plot me dians and variation for each month. Table 4-1: Cross correlations for comparing the three bromeliad data sets (all monitored bromeliads; large class-size T. fasciculata ; and large class-size T. utriculata ) with the average monthly rainfall and the average lowest temperature. Itemized in the table are the lag time; the cross correlation coefficient (CCF); and the alpha value (5%). Average Rain (cm) Average Lowest Temperature (C) Data Set Lag time CCF Alpha (5%) Lag time CCF Alpha (5%) All Bromeliads 0 0.487 0.286 -1 0.645 0.289 Large T. fasciculata -1 0.418 0.289 -1 0.614 0.289 Large T. utriculata -8 0.374 0.312 3 0.332 0.295
35 CHAPTER 5 SURVIVAL OF Tillandsia fasciculata SWARTZ AND Tillandsia utriculata L. IN MYAKKA RIVER STATE PARK Introduction Florida has 16 native species of bromeliads, all of which are epiphytic; 12 of them are susceptible to attack by Metamasius callizona (Chevrolat), an invasive bromeliadeating weevil that escaped into Floridaâ€™s na tural lands in 1989 (F rank and Thomas 1994). Choice of host bromeliad is limited for M. callizona by physiological rest raints; the host plant must have enough biomass to support la rval growth to pupation (Frank and Thomas 1994). Tillandsia usneoides L., T. setacea Swartz, T. bartramii Elliot, and T. recurvata L. are not susceptible to attack by M. callizona because they do not have enough biomass to support larval growth. The other 12 na tive bromeliads are large enough to support larval growth, but only after they have reach ed a certain size. Laboratory research has shown that, for Tillandsia utriculata L., the minimal size susceptible to weevil attack is a plant with longest leaf lengt h of 17.1 +/0.6 cm (Sidoti and Frank 2002). The larger the bromeliad, the more weevil larvae the pl ant can support (Frank and Thomas 1994). Bromeliads have slow growth rates and can take 10-15 y ears to reach maturity and produce seed (Benzing 1980; Isley 1987). Seed s are dispersed by wind, which may carry them to a potential bromeliad host. Germin ation rates are very low for bromeliad seeds (Benzing 1981). Mortality rates are very high for seedlings and small plants and decrease dramatically as the plant becomes larger (Benzing 1980, Isley 1980). Large class-size plants are representative of the reproductive class, and each plant puts out thousands of
36 seeds (Isley 1987); a decline in the reproductive class could have a serious effect on the success of the following genera tion (Frank and Thomas 1994). All of Floridaâ€™s native bromeliads propa gate by seed and by producing vegetative offsets, except for T. utriculata , which dies after going to seed (Isley 1987). The two reproductive strategies result in different life histories and mortality rates. Species that propagate by both seed and vegetative offset s tend to be more stab le and to live longer (Benzing 1980). Because these species split their resources between making seed and supporting the growth of vegetative offsets, they tend to put out less seed per plant; however, because offsets reach maturity much quicker than a plant starting from seed (one to two years versus 10 to 15 years), these species tend to have more regular seed output (Benzing 1980). When T. utriculata goes to seed, the plant dies and leaves no offsets (Isley 1987). This results in a shorter life span, less consistent seed output for a specific location, and a more ephe meral existence (Benzing 1980). The different class-sizes of the bromeliads and the different lif e strategies of the bromeliad species in south Florida could resu lt in different responses to the attack by M. callizona. This chapter examines the effect that M. callizona is having on class-sizes of bromeliads and on two species of bromeliads ( Tillandsia fasciculata Swartz and T. utriculata ) in Myakka River State Park (M RSP) in Sarasota, County, Florida Method and Materials From June 2001 to June 2005, 739 bromeliads were monitored monthly in Myakka River State Park (MRSP). Bromelia d species included in the study were T. fasciculata , T. utriculata , and T. balbisiana Schultes (this was a rare species; only six were found, and two included in the study). The brome liads were classified according to size determined by longest leaf length (see Tabl e C-1 in Appendix C for size classifications
37 based on longest leaf for T. fasciculata and T. utriculata ). Large class-size bromeliads were identified to species. The bromeliads selected for monitoring in MRSP were chosen based on a multitiered method. (See chapter 2 for details on this method and Table A-1 in Appendix A that defines the Tiers). Monitoring was conducted monthly and c onsisted of reloca ting the selected bromeliads and observing whether each plant was alive, dead or missing. Evidence was examined to determine the cause of deaths and disappearances. Parametric survival analysis was used to determine differences in the effect that M. callizona had on the small, medium and large bromeliad class-sizes; and for the large class-size T. fasciculata and large class-size T. utriculata bromeliads. Data were treated as a Weibull distribution and survival cu rves were plotted with a 95% confidence interval. Survival curves were plotted and compared using Log-Rank and Wilcoxon tests for small, medium, and large class-size bromelia ds. Failure for these survival curves was defined as either death or disappearance. Da ta for the small, medium and large class-size bromeliads were then examined using three failure modes: Death by M. callizona ; death by causes other than M. callizona ; and disappearance. Median times to failure for each failure mode were plotted for each classsize, including upper and lower boundaries (+ and Â– 2 standard errors). Survival curves using two modes of failure (death by weevil; and death by causes other than weevil plus disappearances) were plotted for large class-size T. fasciculata and large class-size T. utriculata . The two failure modes for T. fasciculata and T. utriculata were compared using Log-Rank and Wilcoxon test s. Median times to failure for the two
38 modes of failure were tabulated with uppe r and lower boundaries (+ and â€“ 2 standard errors). Results and Discussion Bromeliad seedlings and small plants have very high mortality rates, which decrease dramatically as the plant matu res and becomes larger (Benzing 1980, Isley 1987). The survival plots for the three classsizes of bromeliads in MRSP follow this trend as well. The three curves begin with a high rate of decline that slows down and smoothes out near the end (Fig. 5-1). Steepest rate of dec line is highest for the small class-size, followed by the medium class-size , then the large class-size. Log-rank and Wilcoxon tests have similar Chi square valu es, both significantly higher than the associated P -value (Table 5-1); therefore, the nul l hypothesis, that the curves are the same, was rejected. These survival curves for the small, medi um and large bromeliads (Fig. 5-1) were plotted with failure defined as a death or a disappearance and show the overall pattern for mortality experienced by the bromeliads in MRSP. In order to see how deaths and disappearances vary for the class-sizes, the data for each class-size were analyzed using multiple failure modes. The three failure m odes used were â€˜weevilâ€™ (death caused by M. callizona ); â€˜otherâ€™ (verified deaths, caused by anything except death by M. callizona ); or â€˜disappearanceâ€™ (the plant was mi ssing and could not be relocated). Fig. 52 shows the median time to failure (with uppe r and lower boundaries) for small, medium and large class-sizes for the three modes of failure (weevil, other, and disappear). For small bromeliads, disappearances had a median time to failure of 14.5 months (-1.3 months, + 1.4; n = 250); other deaths had a median time to failure of 56.6 months (12.2 months, + 15.5; n = 53); and weevil deaths had a median time to failure of 124
39 months (56.3 months, + 104; n = 23). Mo st disappearances were probably caused by flooding, storms, high winds, fallen trees and branches, and herbi vory (other than M. callizona , such as from lepidopteran larvae or sq uirrels), and likely resulted in deaths. Other deaths (excluding those caused by M. callizona ) resulted from the same events that caused disappearances; however, the plant rema ins were observed to die from rot, burial, desiccation, cold damage, or herbivory. All of the small bromeliads that were killed by M. callizona were at the high end of the small class-size (approximately 10 to 15 cm, longest leaf length); and, when a M. callizona specimen was found, only one was ever found per plant. This is contrary to laboratory research by Sidoti a nd Frank (2002) where no bromeliads with a longest leaf length less than 17.1 +/0.6 cm supported la rval growth; however, small bromeliad deaths in the field were rare (23 out of 327 deaths and disappearances) among the small class-size bromeliads, and no more than a single larva or empty pupal chamber was discovered in this class-size. For medium-size bromeliads, disappearances had a median time to failure of 37.3 months (4.7 months, + 5.4; n = 70); other de aths had a median time to failure of 87.2 months (36 months, + 60.8; n = 24); and weev il deaths had a median time to failure of 29 months (5 months, + 5.9; n = 31). Causes of disappearances and other deaths were the result of catastroph ic events; other deaths were caused by rot or burial, and, for T. utriculata , going to seed. The median time to failure for disappearances for medium-size plants is similar to that for the small classsize. The median time to failure for weevil deaths for medium class-size plants is similar to that for the large class-size. The wide size range of the medium class-size bromelia ds (15 to 60 cm) moves through a range of
40 morphological characteristics, star ting at the low end with characteristics similar to the small class-size, up to the high end of the range where there is greater similarity to the large class-size. As medium class-size pl ants increased in size and mass, they disappeared less frequently, became less suscep tible to catastrophic events and climatic changes; weevil-killed plants were more of ten found with more than one weevil specimen or empty pupal chamber. For large bromeliads, disappearances had a median time to failure of 113 months (55.6 months, + 110; n = 23); other deaths had a median time to failure of 87.5 months (33.6 months, + 54.5; n = 30); and weevil deaths had a median time to failure of 40.2 months (-7.6 months, + 9.3; n = 51). Disa ppearances were mostly the result of catastrophes, probably caused by storms or fl ood conditions; three disappearances were caused by unlawful collection by humans (all th ree specimens were located near occupied camp grounds, and evidence of knife marks on the bark indicated where the bromeliads had been cut away). Other deaths were cause d by catastrophes resulti ng in rot or burial, and, for T. utriculata , going to seed. Weevil deaths ha d a significantly earlier median time to failure than other deaths or disappear ances. Individual large bromeliads were capable of supporting several weevils. Large bromeliads were identified to species ( T. fasciculata or T. utriculata ) and two failure modes (deaths and disappearance s combined, and weevil deaths) and their affect on large class-size T. fasciculata and T. utriculata populations were examined. Disappearances were assumed to be other deat hs, based on the assumptions that most of the disappearances would result in death; and that these deaths were not caused by M. callizona . These assumptions create a bias towa rds over-estimating the number of other
41 deaths (other deaths plus disappearances ) and under-estimating the number of weevil deaths (assumption that disappe arances all result in other deaths); however, there would be a greater bias in under-es timating the number of other de aths to weevil deaths if disappearances were ignored, because it is a safe assumption that most disappearances resulted in death, not caused by M. callizona ; and death and disappear ance, at the time of its occurrence, effectively removed a poten tial or active reproductive member of the population. Survival curves for the two failure modes for large class-size T. fasciculata have high rates of decline in surviv ability at the beginning, and then smooth out (Fig. 5-3). The survival curve for weevil deaths has a mu ch higher rate of decline in survivability than other deaths and disapp earances. Large class-size T. fasciculata bromeliads have significantly different survival curves for the two failure modes. Log-Rank and Wilcoxon tests have similar Chi-square values that are much higher than their associated P -values (Table 5-2); therefore, the nu ll hypothesis, that the curves are the same, was rejected. The significantly higher rate of decline in the survivability for weevil deaths indicates that mortality caused by M. callizona has a higher probability of happening than death by other causes at any given time. As well, the median time to failure (Table 5-3) for weevil deaths (41 months, ranging from 33 to 51 m onths) is significantly earlier than that for other deaths and disappear ances (83 months, ranging from 50 to 133 months). Survival curves for the two failure modes for large T. utriculata have a high rate of decline in survivability at the beginning, and then smooth out (Fig. 5-4). The survival curve for other deaths and disappearances has a much higher rate of decline in survivability than for weevil deaths. Large T. utriculata have significa ntly different
42 survival curves for the two failure modes. Log-Rank and Wilcoxon tests have similar Chi-square values that are much higher than their associated P -values (Table 5-4); therefore, the null hypothesis , that the curves are the same, was rejected. The significantly higher rate of dec line in the survivability for other deaths indicates that mortality caused by other deaths has a hi gher probability of happening than death by M. callizona at any given time. As well, the median time to failure (Table 5-5) for other deaths and disappearances (14.17 months, rangi ng from 10 to 21 months) is significantly earlier than that for weevil deaths (33 .79 months, ranging from 15 to 58 months). The impact of weevil deaths on the population of large T. fasciculata is greater than the impact of other modes of death; for T. utriculata , other deaths and disappearances have a greater impact than weevil deaths. Both weevil deaths and other modes of death for large T. utriculata have turn over rates that are higher than those for T. fasciculata , as indicated by T. utriculata Â’s steeper survival curves and earlier median times to failure for both failure modes. The population of large T. fasciculata in MRSP was composed of 32% large clumps and 68% individual large plants; larg e clumps and individual plants lost 40 and 38%, respectively, of their population to weevil deaths. Once infestation was apparent, individual plants were kill ed relatively quickly, within a season, much like large T. utriculata plants. Once a weevil infesta tion became apparent on a large T. fasciculata clump, progress of the infestation could take one to three years befo re killing the clump. Clumps continued to seed and make offsets while infested by M. callizona , but the rate of tissue consumption by M. callizona exceeded the rate of an infe sted bromeliadÂ’s ability to regenerate itself.
43 Weevil infestations occurred locally. Wh en one bromeliad on a bromeliad host tree supporting several large T. fasciculata bromeliads became infested, the other plants on the tree would soon become infested, until all or most of the large class-size bromeliads were destroyed. Five Sec tions on the edge of the T. fasciculata population lost all of the large class-size T. fasciculata bromeliads and could not be replaced. Large class-size T. fasciculata bromeliads in the interior suffered lo calized losses on trees and, as mentioned above, the infested trees tended to be located ne ar each other. In the interior, there were both infested and un-infested Sections; most Sections contained one to several large class-size T. fasciculata plants, including clumps. Weevil infestations in a large class-size T. utriculata progressed quickly, within months. During the first 24 m onths of the study, 41 large T. utriculata had been included in the study. Nine of these were killed by M. callizona , 10 went to seed and died, and 8 were washed away in floodwater or rising river water. At the end of 24 months, 13 individuals remained. Over the next two years, eight went to seed and died and two were washed away in rising river water. At the end of the study, only four large plants remained. The overall selected population had de pleted drastically fr om the start to the end of the study, and within the study Sections, there were very few large T. utriculata noticeable. Unlike large T. fasciculata , the loss of large T. utriculata was obvious. In conclusion, M. callizona has a much greater preference for medium and large class-size bromeliads. The two life strategies of T. fasciculata and T. utriculata result in two responses to the attack by M. callizona ; in essence, T. fasciculata , with its large stable populations will have to â€œoutgrowâ€ M. callizona , and T. utriculata , with its more ephemeral life strategy, will have to â€œoutrunâ€ M. callizona .
44 Figure 5-1: Survival plots for small, medium and large class-size bromeliads. Table 5-1: Test statis tics for null hypothesis that all th ree survival curves are the same for small, medium and large class-size bromeliads. Method Chi Square DF P -value Log-Rank 25.6270 2 0.000 Wilcoxon 27.7092 2 0.000
45 Figure 5-2: Median time to failure with upper and lower boundaries , according to size and mode of failure.
46 Figure 5-3: Survival curves for two failure modes (death and disappearance, and weevil) for large class-size Tillandsia fasciculata . Table 5-2: Test statis tics for null hypothesis that the surv ival curves are the same for two modes of death (weevil-kill or other-kill) for large class-size T. fasciculata . Method Chi Square DF P -value Log-Rank 5.63094 1 0.018 Wilcoxon 6.46268 1 0.011 Table 5-3: Median values (months) for failu re modes other deaths and disappearances vs. weevil deaths for large class-size T. fasciculata . Other deaths and disappearances Weevil deaths LB Median UB LB Median UB 50.15 82.72 133.19 32.48 40.61 50.77
47 Figure 5-4: Survival curves for two failure modes (death and disappearance, and weevil) for large class-size Tillandsia utriculata . Table 5-4: Test statis tics for null hypothesis that the surv ival curves are the same for two modes of death (weevil-kill or other-kill) for large class-size T. utriculata . Method Chi Square DF P -value Log-Rank 10.0194 1 0.002 Wilcoxon 6.1293 1 0.013 Table 5-5: Median values (months) for failu re modes other deaths and disappearances vs. weevil deaths for large class-size T. utriculata . Other deaths and disappearances Weevil deaths LB Median UB LB Median UB 9.54 14.17 21.06 14.647 33.79 58.13
48 CHAPTER 6 Metamasius callizona (CHEVROLAT) AND THE FUTURE OF FLORIDAÂ’S NATIVE BROMELIADS Introduction Since Metamasius callizona (Chevrolat) entered Florida and began attacking native bromeliad populations, many people have been wondering what the future outcome of the native bromeliad populations will be in the long term. This thesis has attempted to answer this question to some degree by studying the seasonal trends and patterns exhibited by M. callizona (chapter 3) and some of its hos t bromeliads (chapter 4), and by doing survival analysis on different class-size s of bromeliads and on two species of host bromeliads with different life strategies, Tillandsia fasciculata Swartz and T. utriculata L. (chapter 5). Because of the large sample sizes required to perform seasonal and survival analysis, such analyses were performed only on the monitored bromeliad population in Myakka River State Park, the primary research site. This final chapter will look at the bromeliad populations in the other four Natu ral Areas that were monitored, Loxahatchee National Wildlife Refuge (LNWR), Highl ands Hammock State Park (HHSP), Fakahatchee Strand Preserve St ate Park (FSSP), and St. Seba stian Buffer Preserve State Park (SSSP). Rather than using statistical analysis, this chapter will offer a statistical description of the bromeliad populations in these five Na tural Areas at the e nd of their respective monitoring times. By studying these descri ptions, a few observations can be made
49 concerning general trends and patterns that are notable in the different bromeliad communities and habitats that were represente d in this study. This chapter will conclude by offering some answers to the question of the future outlook for Floridaâ€™s native bromeliads. Methods and Materials Bromeliads were monitored in five Natura l Areas in south Florida: Myakka River State Park, Loxahatchee National Wildlife Refuge, Highlands Hammock State Park, Fakahatchee Strand Preserve State Park, and St. Sebastian Buffer Pr eserve State Park beginning in June 2001 and ending in June 2005. The bromeliads selected for monitoring in the five Natural Areas were chosen based on a multi-tiered method. (See chapter 2 for details on this method and Table A-1 in Appendix A that defines the Tiers). Table B-1 in Appendix B shows the total hectarage mapped for each Natural Area, th e mapping and monitoring schedules, and the bromeliad species that were monitored. The selected bromeliad populations we re monitored monthly; deaths and disappearances were recorded along with an assessment on the cause of death or disappearance. Individual bromeliads were given a unique number and were identified to species (if possible) and categorized according to size. Size was based on longest leaf length; Table C-1 in Appendix C gives the leaf length for the species that were included in this study. The status of the bromeliad population for each Natural Area was broken down according to species and categor ized as alive, killed by M. callizona , killed by causes other than M. callizona , or disappeared. Percentages were plotted as bar graphs for each
50 Natural Area for T. fasciculata , T. utriculata , and Tillandsia balbisiana Schultes; the results for the remaining species were tabulated. The fallen, dead bromeliads that were coll ected monthly from within the Sections were counted for each Natural Area and were categorized as either killed by M. callizona or killed by some other cause based on evidence ( M. callizona specimens found in the plant or other evidence, such as pupal chambe rs, mined inflorescences, chewed interiors, fallen cores if killed by M. callizona ; other deaths included rotting material, injuries, going to seed, etc.). These percentages were compared between the Natural Areas. Results and Discussion From the data collected in MRSP, survival analysis has shown that the reproductive class of the T. fasciculata population suffered greater mortality from M. callizona than from other modes of mortality (primarily catastrophic events th at resulted in rot or burial after the plant fell from its host, or, th e host fell over). The reproductive class of T. utriculata suffered a rate of mortality due to M. callizona less than that caused by other mortality modes (primarily going to seed and, to a lesser extent, catas trophic events that led to rot or burial; see chapter 3). Insufficient data were collected in term s of selected bromeliads and/or time monitored to perform survival analysis for br omeliads at the other four Natural Areas (or from the T. balbisiana population in MRSP that only in cluded two specimens). Figures 6-1, 6-2 and 6-3 show the per centage of the population that remained alive, that was killed by M. callizona , that was killed by a cause other than M. callizona , and that disappeared for T. fasciculata , T. utriculata , and T. balbisiana , respectively, in their associated Natural Areas. Direct comp arisons cannot be made because the time monitored varied and because the populations contained va rying classes-sizes (small,
51 medium and large). Mortality rates vari ed between the class sizes; under normal circumstances, small bromeliads of a given species suffer much higher mortality rates than large bromeliads (Benzing 1980); howev er, medium and large bromeliads suffer greater mortality from M. callizona than do small class-sizes (see chapter 3). Figs. 6-1, 6-2 and 6-3 are included to give a descripti on of the population for each Natural Area at the completion of monitoring, and to point out a few general observations. Table 6-1 shows the outcome for the remaining species that were included in the study in small numbers. For all three species, FSSP had between 70% and 80% of its population remaining alive after 24 months (Figs. 1, 2 and 3). No confirmed M. callizona specimens were found at this study site. One dead medium class-size T. fasciculata had a pupal chamber found in the center, but it was assessed to be Metamasius mosieri Barber, FloridaÂ’s only native bromeliad-eating weevil. One large T. utriculata (the only death in the T. utriculata population) died from going to seed. The remaining deaths and disappearances for all three specimens included only small a nd medium class-size bromeliads and were caused by catastrophic events. In MRSP, at the end of 49 months, the T. fasciculata population had lost 58% of its population, of which 36.5% were killed by M. callizona (Fig. 6-1). Losses to the population were obvious along the edge of the population and in localized spots within the population. However, 42% of the populatio n remained alive and, throughout the park, many large T. fasciculata plants were noticeable. Monitored bromeliads included only large (and some medium) class-size T. fasciculata because it was difficult to make an accurate identification betw een the medium and small T. fasciculata and T. utriculata .
52 Monitored T. fasciculata at HHSP consisted of five large plants and one medium. After 33 months, the medium plant and one large plant remained alive. Metamasius callizona killed one large plant and the remaini ng three large plants disappeared due to branch falls and parks personne l doing maintenance work (cleaning up the fallen debris). LNWR and SSSP both had roughly 60% of their monitored T. fasciculata alive after 28 and 17 months, respectively; bo th populations included all class-sizes. LNWR had a higher percentage of plants killed by M. callizona than by other deaths, and a lower percentage of disappearances compared to that experienced in SSSP (20% compared to 35%, respectively). SSSP also had a very low percentage of T. fasciculata plants killed by M. callizona , especially for an area that had an active infestation ongoing; weevil specimens and fallen dead bromeliads w ith evidence of weevil-kill were found consistently throughout the study, with the excep tion of the last three months. The high percentage of disappearances was caused pr imarily by hurricane activity in September 2004; 27 plants (10 small, 6 medium and 11 la rge) were removed during the storms. Six of the large plants showed possible signs of weevil infestation and this may have contributed to their dislodgeme nt from their host. Both LNWR and SSSP had plants with signs of infestation by M. callizona in the remaining population. In MRSP, at the end of 49 months, the T. utriculata population had lost 89% of its population, of which 19% were killed by M. callizona (Fig. 6-2). Other deaths (46%) were caused by going to seed, except for one plant. Disappearances (34%) were the result of catastrophic events. HHSP had a si milar pattern, with a higher percentage of deaths due to M. callizona and fewer deaths and disappe arances. FSSP, like HHSP, had a
53 small population of T. utriculata , and, as mentioned above, the one death was due to seed production and the disappearance was the result of a catastrophic event. In MRSP, at the end of 49 months, one of the two T. balbisiana plants died from rot and the other disappeared in flood waters (Fig. 6-3). A small percentage (5%) of T. balbisiana plants in LNWR was killed by M. callizona ; none were killed by M. callizona in SSSP (though two T. balbisiana were killed by M. mosieri ). Of the remaining species ( Tillandsia simulata Small, T. variabilis Schlechtendal, T. paucifolia Baker, T. pruinosa Swartz, and Guzmania monostachia L.), most of the populations were relatively stab le (Table 6-1). Only one T. paucifolia at SSSP was killed by M. callizona . Both T. paucifolia and T. pruinosa lost six of their members to catastrophic events. One G. monostachia plant was killed by ex cessive foliage feeding by the fall webworm, Hyphantria cunea (Drury) (Lepidoptera: Arctiidae). Table 6-2 shows the number of dead brome liads that fell from the canopy and that were collected from each Natural Area. Six hundred fifty-two dead bromeliads fell from the canopy into the study Sections in MRSP; 78% were killed by M. callizona and 22% were killed by other factors. In LNWR, 93 dead bromeliads were collected; 71% were killed by M. callizona . In HHSP, 31 dead bromeliads we re collected; 77% were killed by M. callizona . The percentages of dead bromeliads collect ed from the ground that were killed by M. callizona were rather consistent (within 11%) considering the different habitats and bromeliad communities from which they were collected. As well, they were relatively high (ranging from 71 to 82%), much highe r than deaths observed from selected bromeliads in the canopy. This is because death caused by M. callizona usually results in
54 part of the infested plant falling to the ground due to internal chewing by the weevil larva(e) inside the plant; and because most infested plants ar e medium or large class-size and the fallen mass is large enough to persist. Roughly half (66 of 129) of the dead brom eliads collected in SSSP were discovered on the initial monitoring trip in the oak hammock. All 66 were dead T. utriculata plants, and ranged from old, brown and dried up plants, to plants with soft tissue remaining in the center. They all showed signs of weevil damage and some contained pupal chambers (25 total were collected from the 66 plants) a nd live weevils (two adults and one larva). There were no remaining T. utriculata in the canopy within the demarcated Sections, only T. fasciculata . Only a small portion of the hammock was mapped (0.04 hectares); the T. utriculata populating the rest of the hammock had suffered the same fate and littered the ground. Of all the bromeliads so far observed in the field, T. utriculata has apparently suffered the most. Situations such as th at in the oak hammock in SSSP, the loss of the T. utriculata population in LNWR, and the obvious decline in MRSP and HHSP are not isolated events. Devastated T. utriculata populations have frequently been observed in south Florida (Frank 1996b). The large and medi um class-size plants are being affected the greatest (Chapter 3) and their demise is interfering with T. utriculata Â’s seed production. The slow growth rate (10 to 15 years) to maturity (Benzing 1980, Isley 1987) and the reliance of T. utriculata on seed production to supply the following generation have profound effects on the future of this species. Because T. utriculata has a patchy distribution and an ephemeral na ture, it may escape infestation from M. callizona if small patches can reach maturity and go to seed before discovery by M.
55 callizona . However, once a patch is discovered, M. callizona can destroy it in a year to a few years. Large patches (such as the patc h that once existed in the oak hammock in SSSP) are especially vulnerable because they are more easil y found, and just as easily destroyed; and because such large patches ar e more stable, such losses represent the loss of seed reservoirs and can have profound effects on the continuation of the species. Tillandsia fasciculata is a more stable bromeliad species that maintains its populations through the production of vegetative offsets as well as by seed production (Benzing 1980, Isley 1987). Individual pl ants, as they mature and vegetatively propagate, become giant clumps. Metamasius callizona may take one to three years to kill an individual clump, and even longer to destroy a population (see chapter 3). It is difficult to assess the long-term outcome of M. callizona Â’s attack on T. fasciculata . The process is very slow, but the rate of mortality that is inflicted on the reproductive class is quite high (see chapter 3), and over time, could prove fatal to the species. A more likely outcome is that the population will become greatly reduced, yet be able to maintain a stable M. callizona population, from which M. callizona could continue to persist in the natural lands, while continuing to spread out and attack other susceptible bromeliad populations, such as those in the Everglades. Metamasius callizona has been sighted only once, in March 2002, in FSSP, and has since been spotted several times in the Big Cypress National Pr eserve (Frank 1996b). That M. callizona was not found in the study site in FSSP demonstrates only that M. callizona has not yet found that very small area w ithin the greater area. FSSP, as well as the Everglades in general, has the highest di versity of FloridaÂ’s native bromeliads; most are patchily distributed, and many are rare (Benzing 1980). The arrival of M. callizona
56 into this area heralds the potential for the loss of numerous species of bromeliads (Frank and Cave 2005). Part of the difficulty in determining th e future outcome of FloridaÂ’s native bromeliads in response to M. callizona is caused by the number ( 12) of species that are affected, and the different habitats and life stra tegies these species have. Factors that will determine the long-term survivab ility for a particular species of bromeliad include 1) the increased rate of mortality caused by M. callizona in relation to the speciesÂ’ ability to outgrow or outrun M. callizona attack, 2) the range and rarity of the species, and 3) the distribution of the bromeliad patches. Survival analysis has shown that M. callizona has a great effect on the mortality rate of T. fasciculata and T. utriculata , but in different ways. The weevil consumes T. fasciculata slowly, and the infestation is a long pr ocess because of the large biomass and quick growth rate of the vegetative offsets produced by T. fasciculata , and because T. fasciculata has coriaceous leaves that would be hard to consume and digest (Benzing 1980). In order to survive, T. fasciculata must outgrow the weevil. Tillandsia utriculata , because it has an ephemeral life style and dies upon going to seed, must outrun the weevil. Infestations move quickly in a patch of Tillandsia utriculata because, despite the large body size, the plant has soft, easily c onsumed leaves and produces no offsets to replace eaten tissue. It is possible that M. callizona exhibits seasonal p opulation increases on T. utriculata (in the spring; see chapter 3). Th is could complicate understanding how M. callizona would shift geographically, and how the distribution of brom eliad patches may affect the survivability of some of the patches. While large T. fasciculata populations
57 may support a long-term population of M. callizona in a geographic spot, as T. utriculata patches move from one location to another, small, localized, and e phemeral outbreaks of M. callizona could arise near other bromeliad patches. This thesis has only examined T. fasciculata and T. utriculata in great detail, and one can only speculate on how the other specie s will respond to the weevil infestation. Guzmania monostachia (Larson 2000c) is a large br omeliad that reproduces both vegetatively and by seed; its range is restricted to the lower tip of the peninsula, but large populations can be found. I ndividual plants grow as giant clumps, much like T. fasciculata ; however, G. monostachia has soft leaves, softer than T. utriculata . Guzmania monostachia , with such large populations, may not be able to outrun M. callizona ; will it be able to outgrow the weevil, or will the we evil be able to quickly consume the plantÂ’s soft tissue? Tillandsia pruinosa is a very rare plant with a range limited to Collier County in the southern end of the peninsula (Larson 2000c). It is a small plant, only reaching about 25 cm long and it has tough leaves; it grows in small patches both by vegetative offsets and by seed. It is unknown how this pl ant will react to an attack by M. callizona ; however, the very rarity of the species and the small patches make this plant vulnerable to any added stresses. Metamasius callizona has not been observed attacking M. pruinosa (Frank and Cave 2005), however it is a safe assumption that T. pruinosa would be susceptible to attack by M. callizona because the plant has enough biomass to support larval growth and because M. callizona has been observed attacking a similar, wider ranging species, T. paucifolia .
58 There are several other bromeliads that ar e rare and that grow in small patches, including three Catopsis species and T. variabilis ; as well as some with wider ranges, including T. simulata , FloridaÂ’s only precinctive species of bromeliad (Larson 2000c). Canopy scientists have broadened our understanding of canopy ecosystems, and biological control has increased our understandi ng of invasive species, but there is still much to learn about both of these topics. The lack of information available on bromeliads in general, on FloridaÂ’s brome liads in particular, and on bromeliad-eating weevils is pronounced. The weevil infestation in Florida is unfort unate; however, it has opened an opportunity and a need to learn mo re about FloridaÂ’s native bromeliads and about bromeliad-eating weevils. The more urge nt tasks are to continue research on the potential biological control agent, the tachinid fly Lixadmontia franki Wood and Cave, which is in quarantine at the UF-IFAS Bi ological Control Research and Containment Laboratory in Ft. Pierce. The objective is to rear the fly in great enough quantity that it can be released into natural areas, become established, and reduce the M. callizona population by parasitizing the weevil larvae. Post-monitoring will be necessary to know the results of this process.
59 Tillandsia fasciculata Status at Completion of Study MRSP, LNWR, FSSP, SSSP 0 10 20 30 40 50 60 70 80 MRSP; time = 49 mos. LNWR; time = 28 mos. HHSP; time = 33 mos. FSSP; time = 24 mos. SSSP; time = 17 mos. Natural AreaPercent per Category alive M. callizona other disappear Figure 6-1: Status of Tillandsia fasciculata bromeliads (given as a percentage in the following categories: Alive; killed by M. callizona ; killed by cause other than M. callizona ; or disappeared) in MRSP, LNWR, HHSP, FSSP, and SSSP. Tillandsia utriculata S tatus at Completion of Study MRSP, HHSP, FSSP 0 10 20 30 40 50 60 70 80 90 MRSP; time = 49 mos. HHSP; time = 33 mos. FSSP; time = 24 mos. Natural AreaPercent per Category alive M. callizona other disappear Figure 6-2: Status of Tillandsia utriculata bromeliads (given as a percentage in the following categories: Alive; killed by M. callizona ; killed by cause other than M. callizona ; or disappeared) in MRSP, HHSP, and FSSP.
60 Tillandsia balbisiana Status at Completion of Study MRSP, LNWR, FSSP, SSSP 0 20 40 60 80 100 MRSP; time = 49 mos. LNWR; time = 28 mos. FSSP; time = 24 mos. SSSP; time = 17 mos. Natural AreaPercent per Category alive M. callizona other disappear Figure 6-3: Status of Tillandsia balbisiana bromeliads (given as a percentage in the following categories: Alive; killed by M. callizona ; killed by cause other than M. callizona ; or disappeared) in MRSP, LNWR, FSSP, and SSSP. Table 6-1: Other species of bromelia ds monitored in HHSP, FSSP and SSSP and condition at end of monitoring period. Species HHSP (at 33 months) FSSP (at 24 months) SSSP (at 17 months) Tillandsia simulata 2 (alive) -2 (alive) Tillandsia variabilis -2 (alive) -Tillandsia paucifolia --6 (alive) 1 ( M. callizona ) 6 (disappear) Tillandsia pruinosa -16 (alive) 6 (disappear) -Guzmania monostachia -10 (alive) 1 (other) -
61 Table 6-2: Total number of dead bromelia ds fallen from canopy into mapped Sections, collected, and categorized as percent killed by M. callizona or killed by cause other than M. callizona . Natural Area Months monitored Total Number Fallout % M. callizona % other MRSP 49 652 78% 22% LNWR 28 93 71% 29% HHSP 33 31 77% 23% FSSP 24 9 0% 100% SSSP 17 158 82% 18%
62 APPENDIX A METHOD TIERS Table A-1: Description and parameters for the five tiers used to define demarcated Sections and Bromeliad Hosts. Tier Description Parameters I Natural Areas State and Federal Parks and Refuges. Bromeliad-supporting habitat present. II Region Bromeliad-supporting habitat in the Natural Areas. At least 10 bromeliad hosts must make up this tier at the initial mapping; there is no upper limit. III Area Regions were divided into Areas based on bromeliad host density and local landmarks that defined the Area. A Region could have 1 or more Areas; upper limit was defined by the habitat, or by limitations in resources and time. 10 Â– 100 bromeliad hosts per Area must be present in an area at the initial mapping. IV Section Areas were divided into sections based on bromeliad host density. An Area could have 3 to 10 Sections. Of these Sections, half (or half of the Sections + 0.5, if there was an odd number of Sections) were randomly selected for monitoring Â– collecting weevil specimens from dead bromeliads monthly. 3-10 bromeliad hosts per Section must be present in a Section at the initial mapping. V Bromeliad host Each Section contained 3 to 10 bromeliad hosts; of these bromeliad hosts, half (or half of the bromeliad hosts + 0.5, if there was an odd number of Sections) were randomly selected for monitoring. Bromeliad hosts were sketched from a particular direction and the bromeliads growing on the host were indicated on the sketch; these are the bromeliads that were monitored monthly. Bromeliads included in the sketch were selected based on apparency, that is, those that were most obvious. If the bromeliad host were covered with numerous bromeliads, then up to 15 were included. If these bromeliads died or disappeared, then the bromeliad host would be examined for any other bromeliads; if any were found, they replaced th e lost bromeliads and were monitored; if no other bromeliads could be found on the host, then another bromeliad host in the Section was randomly selected for monitoring. Updates to replace lost or dead bromeliads were made every six months.
63 APPENDIX B MAPPING SCHEDULE Table B-1: Total hectarage mapped for each Natural Area, mapping and monitoring schedules, and bromeliad species suscepti ble to weevil attack present for each of the five Natural Areas. Natural Area Hectares mapped Mapping schedule Bromeliad species MRSP 2.98 June 2001 Â– Initial mapping; began monitoring. December 2001 Â– Adde d five ne w Areas. February 2002 Â– Added 1 new Region and 1 new Area. June 2005 Â– Ended study. Tillandsia fasciculata Swartz Tillandsia utriculata Linnaeus Tillandsia balbisiana Schultes LNWR 0.05 April 2002 Â– Initial mapping; begin monitoring. March 2003 Â– Added 1 new Region. February 2005 Â– Ended monitoring. Tillandsia fasciculata Swartz Tillandsia balbisiana Schultes HHSP 0.45 August 2002 Â– Initial mapping; began monitoring. September 2002 Â– Added 1 new Region. April 2005 Â– Ended monitoring. Tillandsia fasciculata Swartz Tillandsia utriculata Linnaeus Tillandsia balbisiana Schultes Tillandsia simulata Small FSSP 0.04 March 2003 Â– Initial mapping; began monitoring. March 2005 Â– Ended monitoring Tillandsia fasciculata Swartz Tillandsia utriculata Linnaeus Tillandsia balbisiana Schultes Tillandsia variabilis Schlechtendal Guzmania monostachia (Linnaeus) Tillandsia pruinosa Swartz SSSP 0.57 November 2003 Initial mapping; began monitoring. March 2005 Â– Ended monitoring. Tillandsia fasciculata Swartz Tillandsia utriculata Linnaeus Tillandsia balbisiana Schultes Tillandsia paucifolia Baker
64 APPENDIX C LONGEST LEAF LENGTH Table C-1: Size classificatio ns, defined by longest leaf le ngth, for the bromeliad species in this study. Longest Leaf Size Categories (cm) Species Small Medium Large Tillandsia fasciculata 15 15-60 > 60 Tillandsia utriculata 15 15-60 > 60 Tillandsia balbisiana 5 5-15 > 15 Tillandsia simulata 5 5-15 > 15 Tillandsia variabilis 10 10-30 > 30 Guzmania monostachia 15 15-60 > 60 Tillandsia paucifolia 2.5 2.5-5 > 5 Tillandsia pruinosa 2.5 2.5-5 > 5
65 APPENDIX D HEALTH RATINGS Table D-1: Health rating chart showing the cl assifications for the four quarters of the physiological condition scale. Second quarter (moderately stressed): Outward stresses are more apparent; leaves chewed, moderate discoloration; slight de siccation or freeze trauma. If the stresses were removed, the plant would likely recover. First quarter (healthy): At 3.0, the bromeliad has good, strong green color and no discoloration; no obvious injuries; and turgid leaves. As the health rating falls to 2.5, there may be a few injuries or minor discoloration, but nothing serious. Third quarter (heavily stressed): Heavy injury, major discoloration or browning of leaves covering up to half of the plantÂ’s biomass; core leaves falling out; severe frost or drought damage. If the stress or injuries were removed, the plant might recover , but would suffer lon g -term conse q uences from the ex p erience. Fourth quarter (seriously stressed): Very poor health; heavy injury and/or loss of leaves affecting more than half of the plantÂ’s biomass; core missing; severe discoloration. Death = 1.0 = no green tissue on plant.
66 LIST OF REFERENCES Benzing DH. 1980. The Biology of the Bromeliads. Mad River Press, Eureka, California. Benzing DH. 1991. Epiphytic bromeliads as air quality monitors in south Florida. Selbyana 12:46-53. Butler J. 1974. Pineapples of the treetops. Florida Nat 47(4):13-17. Cave RD. 1997. Admontia sp., a potential biolog ical control agent of Metamasius callizona . J Brom Soc 47:244-249. Ferriter A. 2006. Distribution of Mexican brom eliad weevil in south and central Florida. Fig. 9-3 In: 2006 South Florida Environmen tal Report. Florida Department of Environmental Protection (DEP) and Sout h Florida Water Management District (SFWMD). Frank JH. 1983. Bromeliad phytotelmata and th eir biota, especially mosquitoes. In: Frank JH, Lounibos LP. Phytotelmata: terrest rial plants as hosts for aquatic insect communities. Plexus, Medford, New Jersey. P 101-128. Frank JH. 1996a. Bromeliad biota: Biology of the weevil Metamasius callizona . http://BromeliadBiota .ifas.ufl.edu/wvbrom5.htm. (June 2006). Frank JH. 1996b. Bromelia d biota: History of Metamasius callizona in Florida. http://BromeliadBiota .ifas.ufl.edu/wvbrom6.htm. (June 2006). Frank JH. 1999. Bromeliad-eati ng weevils. Selbyana 20:40-48. Frank JH, Cave RD. 2005. Metamasius callizona is destroying FloridaÂ’s native bromeliads. In: Second Internationa l Symposium on Biological Control of Arthropods; 2005 Sep12-16; Davos, Switzer land: USDA Forest Service Publication FHTET-2005-08. Vol 1. P 91-101. Frank JH, Cooper TM, Larson BC. 2006. Metamasius callizona (Coleoptera: Dryopthoridae): Longevity and fecundity in the laboratory. Florida Entomol 89:208-211. Frank JH, Curtis GA. 1981. Bionomics of the bromeliad-inhabiting mosquito Wyeomyia vanduzeei and its nursery plant Tillandsia utriculata . Florida Entomol 64:491-506.
67 Frank JH, OÂ’Meara GF. 1985. Influence of microand macrohabitat on distribution of some bromeliad-inhabiting mosqu itoes. Entomol Exp Appl 37:169-174. Frank JH, Sreenivasan S, Benshoff PJ, Deyrup MA, Edwards GB, Halbert SE, Hamon AB, Lowman MD, Mockford EL, Scheffrahn RH, Steck GJ, Thomas MC, Walker TJ, Welbourn WC. 2004. Invertebra te animals extracted from native Tillandsia (Bromeliales: Bromeliaceae) in Sarasota County, Florida. Florida Entomol 87:176185. Frank JH, Thomas MC. 1994. Metamasius callizona (Chevrolat) (Coleoptera: Curculionidae), an immigrant pest, destroys bromeliads in Florida. Can Entomol 126:673-682. Isley PT. 1987. Tillandsia the worldÂ’s most unusual air plants. Botanical Press, Gardena, California. Larson BC. 2000a. Save FloridaÂ’s native br omeliads: Damage caused by the Mexican bromeliad weevil. http://savebromeliad s.ifas.ufl.edu/damage.htm. (June 2006). Larson BC. 2000b. Save FloridaÂ’s native brome liads: Biological cont rol of the Mexican bromeliad weevil. http://savebromeliads. ifas.ufl.edu/biocontrol.htm. (June 2006). Larson BC. 2000c. Save FloridaÂ’s nativ e bromeliads: FloridaÂ’s bromeliads. http://savebromeliads.ifas.ufl.e du/bromeliads.htm. (June 2006). Larson BC. 2000d. Save FloridaÂ’s native bromelia ds: Value of bromelia ds to the state. http://savebromeliads.ifas.ufl.edu/value.htm. (June 2006). Lopez LCS, Rios RI. 2001. Phytotelmata faunal communities in sun-exposed versus shaded terrestrial bromeliads from sout heastern Brazil. Selbyana 22(2):219-224. McNeely, J.A. 1998. The science of the canopy: who needs the information and how they can use it. Selbyana 19(2):135-140. Minitab, Inc. 2006. Minitab. http://minitab.com. (June 2006). Myers RL, Ewel JJ. 1990. Ecosystems of Fl orida. University of Central Florida, Orlando, Florida. Nadkarni NM. 1992. The conservation of epip hytes and their habi tats: summary of a discussion at the international sympos ium on the biology and conservation of epiphytes. Selbyana 13:140-142. National Weather Service. 2005. Nationa l Weather Service, Southern Region Headquarters: Climate Â– Past weather. http://www.srh.noaa.gov/tbw/html/tbw/clim ate/climatemain.htm. (June 2006).
68 Salas J, Frank JH. 2001. Development of Metamasius callizona (Coleoptera: Curculionidae) on pineapple st ems. Florida Entomol 84:123-126. Sidoti BJ, Frank JH. 2002. The e ffect of size of host plant ( Tillandsia utriculata : Bromeliaceae) on development of Metamasius callizona (Dryophthoridae). Selbyana 23(2):220-223. Wood DM, Cave RD. 2006. Description of a new genus and species of weevil parasitoid from Honduras (Diptera: Tachinidae). Florida Entomol 89:239-244.
69 BIOGRAPHICAL SKETCH Teresa M. Cooper was born in Rhode Island, USA, in 1966. She attended Eastern Kentucky University (Richmond, Kentucky) from September 1990 to December 1992. She moved to Florida in the spring of 1993 and graduated from Santa Fe Community College (Gainesville, Florida) in 1995 with an Associate of Science in environmental science. In 2002, she received her Bachelor of Science in entomo logy, specializing in plant protection, from the University of Fl orida (Gainesville, Florida). During her undergraduate years, she starte d field research monitoring an invasive bromeliad-eating weevil that was attacking FloridaÂ’s native brom eliads; she carried this research into her graduate studies at the Entomology and Nema tology Department in the University of Florida, and it has culminated in this thesis.