<%BANNER%>

Firefly Diversity in Colombia

Permanent Link: http://ufdc.ufl.edu/UFE0024425/00001

Material Information

Title: Firefly Diversity in Colombia Patterns across a Dynamic Landscape
Physical Description: 1 online resource (123 p.)
Language: english
Creator: Smith, Bradley
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: colombia, lampyridae, maxent, modeling, neotropics
Entomology and Nematology -- Dissertations, Academic -- UF
Genre: Entomology and Nematology thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: The diversity of Neotropical Lampyridae has not been addressed since the 19th century. The charismatic nature of fireflies contributes to their use as conservation units, especially in the Neotropical region where the group exhibits much diversity. The diversity of fireflies within the country of Colombia, with two designated hotspots of worldwide biodiversity, is herein quantified for the first time across both spatial and temporal scales. The population distribution modeling software ?MaxEnt? was used to estimate the distribution of Lampyridae across Colombia as well as the distribution of individual species where possible. Fisher?s alpha, Evenness Values, and Species Accumulation Curves were calculated via the software package ?EstimateS? for the collection sites used in this study. Firefly abundance and species richness values were calculated for each month throughout the study. Differences between the geographical distributions of the different sexual signaling systems (pheromone, photic + pheromone, photic) were modeled using ?MaxEnt.? Data collected over the course of this study indicates that firefly species are not distributed randomly across habitats in Colombia and that the estimated number of firefly species still awaiting capture within Colombia is greater than the total amount of firefly species collected over the course of this study. In addition, each sexual signaling system was also not distributed in a random manner. Finally, data from this study suggests that there are differences between the geographic distributions of the each firefly sexual signaling systems. Hopefully the patterns of species distributions across time and space reported here will be of value to future researchers in their attempts to study and conserve biodiversity in the Neotropics in general, and the country of Colombia in particular.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Bradley Smith.
Thesis: Thesis (M.S.)--University of Florida, 2009.
Local: Adviser: Branham, Marc A.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-05-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024425:00001

Permanent Link: http://ufdc.ufl.edu/UFE0024425/00001

Material Information

Title: Firefly Diversity in Colombia Patterns across a Dynamic Landscape
Physical Description: 1 online resource (123 p.)
Language: english
Creator: Smith, Bradley
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: colombia, lampyridae, maxent, modeling, neotropics
Entomology and Nematology -- Dissertations, Academic -- UF
Genre: Entomology and Nematology thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: The diversity of Neotropical Lampyridae has not been addressed since the 19th century. The charismatic nature of fireflies contributes to their use as conservation units, especially in the Neotropical region where the group exhibits much diversity. The diversity of fireflies within the country of Colombia, with two designated hotspots of worldwide biodiversity, is herein quantified for the first time across both spatial and temporal scales. The population distribution modeling software ?MaxEnt? was used to estimate the distribution of Lampyridae across Colombia as well as the distribution of individual species where possible. Fisher?s alpha, Evenness Values, and Species Accumulation Curves were calculated via the software package ?EstimateS? for the collection sites used in this study. Firefly abundance and species richness values were calculated for each month throughout the study. Differences between the geographical distributions of the different sexual signaling systems (pheromone, photic + pheromone, photic) were modeled using ?MaxEnt.? Data collected over the course of this study indicates that firefly species are not distributed randomly across habitats in Colombia and that the estimated number of firefly species still awaiting capture within Colombia is greater than the total amount of firefly species collected over the course of this study. In addition, each sexual signaling system was also not distributed in a random manner. Finally, data from this study suggests that there are differences between the geographic distributions of the each firefly sexual signaling systems. Hopefully the patterns of species distributions across time and space reported here will be of value to future researchers in their attempts to study and conserve biodiversity in the Neotropics in general, and the country of Colombia in particular.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Bradley Smith.
Thesis: Thesis (M.S.)--University of Florida, 2009.
Local: Adviser: Branham, Marc A.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-05-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024425:00001


This item has the following downloads:


Full Text

PAGE 1

1 FIREFLY DIVERSITY IN COLOMBIA: PATTERNS ACROSS A DYNAMIC LANDSCAPE By BRADLEY WILLIAMS SMITH 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 2009

PAGE 2

2 2009 Bradley Williams Smith

PAGE 3

3 To my father, who starte d me on this long journey

PAGE 4

4 ACKNOWLEDGMENTS I would like to thank everyone who has encouraged and helped m e in my entomological studies, especially Dr. Keith Philips for introduc ing entomology to me, Kyle Beucke for helping me with many of my systematic questions and discussing every nuance of the subject with me, and Dr. Marc Branham for taking me on as his student and helping me with the frustrations of distribution modeling programs and id entifying insects that have not been systematically studied in many decades. I thank also Dr. Skip Choate and Dr. Jaret Daniels for their support of my studies. I also thank my wife, Sheri, who has supported me through all of our time together, and my mother and sister who have believed in me ev ery step of the way despite my failures. Finally, I would like to thank Dr. Michae l Sharkey of the University of Kentucky for providing the specimens used in this study.

PAGE 5

5 TABLE OF CONTENTS Page ACKNOWLEDGMENTS ............................................................................................................... 4LIST OF TABLES ...........................................................................................................................7LIST OF FIGURES .........................................................................................................................8LIST OF TERMS/SYMBOLS/ABBREVIATIONS .....................................................................11ABSTRACT ...................................................................................................................... .............12CHAPTER 1 INTRODUCTION .................................................................................................................. 142 LITERATURE REVIEW .......................................................................................................173 FIREFLY DISTRIBUTIONS AND DIVERS IT Y ACROSS THE GEOGRAPHIC RANGE OF COLOMBIA ...................................................................................................... 26Introduction .................................................................................................................. ...........26Materials and Methods ...........................................................................................................27Specimens ..................................................................................................................... ...27Distribution Modeling ..................................................................................................... 28Results .....................................................................................................................................37Distribution Models ......................................................................................................... 37Diversity Estimation ........................................................................................................ 37Discussion .................................................................................................................... ...........394 FIREFLY ABUNDANCE AND SPECIES RICHNESS ACR OSS ELEVATION AND THROUGH TIME ..................................................................................................................46Introduction .................................................................................................................. ...........46Materials and Methods ...........................................................................................................49Specimens ..................................................................................................................... ...49Statistical Methods ..........................................................................................................49Results .....................................................................................................................................50Discussion .................................................................................................................... ...........50

PAGE 6

6 5 GEOGRAPHICAL DISTRIBUTION OF COLOMBIAN FIREFLY SEXUAL SIGNALING MODALITIES ................................................................................................. 60Introduction .................................................................................................................. ...........60Materials and Methods ...........................................................................................................61Specimens ..................................................................................................................... ...61Determining Sexual Signaling System ............................................................................ 62Distribution Modeling ..................................................................................................... 62Results .....................................................................................................................................63Distribution of Morphospecies and Indi viduals in Different Sexual Signaling Modalities .................................................................................................................... 63Distribution Models ......................................................................................................... 63Discussion .................................................................................................................... ...........686 CONCLUSIONS ................................................................................................................... .73Objective 1 ..............................................................................................................................73Objective 2 ..............................................................................................................................74Objective 3 ..............................................................................................................................75Comparisons and Concluding Remarks ..................................................................................76APPENDIX A MAXENT MORPHOSPECIES DISTRIBUTION MAPS .................................................. 78B SPECIES ACCUMULATION CU RVES FOR SITES REPR ESENTED BY FIREFLY SPECIMENS ..................................................................................................................... .....97C TABLE OF SPECIES PRES ENCE I N EACH SITE ........................................................... 107LIST OF REFERENCES .............................................................................................................114BIOGRAPHICAL SKETCH .......................................................................................................123

PAGE 7

7 LIST OF TABLES Table Page 3-1 Sites within Colombia represente d by collected firefly specim ens ................................... 29 3-2 WorldCLIM Data layers used in analysis .......................................................................... 33 5-1 Number of morphospecies and individuals by sexual signaling system type .................... 64 C-1 Presence and absence of morphospecies by site .............................................................. 108

PAGE 8

8 LIST OF FIGURES Figure Page 3-1 Map of Colombia showing all collection sites (not all sites yielded fireflies) ................... 30 3-2 Log-likelihood map of pred icted firefly distribution for Colombia and imm ediately surrounding areas ............................................................................................................. ..38 3-3 Morphospecies and indivi duals caught across parks ......................................................... 40 3-4 Fishers values and Evenness across parks ..................................................................... 41 3-5 Pooled species accumulation curves .................................................................................. 42 4-1 Morphospecies and individuals caugh t through the course of the study ........................... 51 4-2 Morphospecies and individuals caugh t across elevations of Colom bia ............................. 52 4-3 Morphospecies and individuals caught th rough the course of the study with the ONI index plotted.......................................................................................................................54 4-4 Morphospecies and individua ls caught through the course of the study with the tim e period of rainy seasons superimposed ............................................................................... 56 4-5 Morphospecies and individuals caught through the course of the study with the num ber of traps plotted ......................................................................................................57 4-6 Morphospecies and individua ls caught and m alaise traps used across elevations of Colombia with the number of traps plotted .......................................................................59 5-1 Log-likelihood map of pr edicted pherom one signaling morphospecies distribution for Colombia and immediately surrounding areas ............................................................. 65 5-2 Log-likelihood map of predicted photic + pheromone signaling m orphospecies distribution for Colombia and immediately surrounding areas ......................................... 66 5-3 Log-likelihood map of pr edicted photic signaling m or phospecies distribution for Colombia and immediately surrounding areas .................................................................. 67 5-4 Log-likelihood distributions for sexual sign aling modalities for areas surrounding the Amazon River ....................................................................................................................70 5-5 Distribution of number of species with dif ferent sexual signaling modalities across elevation. .................................................................................................................... ........72 A-1 Log-likelihood map of pr edicted distributions for se lected m orphospecies of Lucidota in Colombia and immedi ately surrounding areas ............................................... 79

PAGE 9

9 A-2 Log-likelihood map of pr edicted distributions for se lected m orphospecies of Lucidota in Colombia and immedi ately surrounding areas ............................................... 80 A-3 Log-likelihood map of pr edicted distributions for se lected m orphospecies of Lucidota Ledocas and Macrolampis in Colombia and immediately surrounding areas ......................................................................................................................... ..........81 A-4 Log-likelihood map of pr edicted distributions for se lected m orphospecies of Photuris in Colombia and immediately surrounding areas .............................................................. 82 A-5 Log-likelihood map of pr edicted distributions for se lected m orphospecies of Photinus and Phaenolis in Colombia and immediately surrounding areas ....................... 83 A-6 Log-likelihood map of pr edicted distributions for se lected m orphospecies of Dryptelytra Pygolampis and Tenaspis in Colombia and immediately surrounding areas ......................................................................................................................... ..........84 A-7 Log-likelihood map of pr edicted distributions for se lected m orphospecies of Magnoculus in Colombia and immediately surrounding areas ..........................................85 A-8 Log-likelihood map of pr edicted distributions for se lected m orphospecies of Magnoculus and Psilocladus in Colombia and immediately surrounding areas ............... 86 A-9 Log-likelihood map of pr edicted distributions for se lected m orphospecies of Psilocladus in Colombia and immediately surrounding areas ........................................... 87 A-10 Log-likelihood map of pr edicted distributions for se lected m orphospecies of Psilocladus and unidentified morphospecies in Colombia and immediately surrounding areas ............................................................................................................. ..88 A-11 Log-likelihood map of predicted distributions for selected unidentified morphospecies in Colom bia and immediately surrounding areas ..................................... 89 A-12 Log-likelihood map of predicted distributions for selected unidentified morphospecies in Colom bia and immediately surrounding areas ..................................... 90 A-13 Log-likelihood map of predicted distributions for selected unidentified morphospecies in Colom bia and immediately surrounding areas ..................................... 91 A-14 Log-likelihood map of predicted distributions for selected unidentified morphospecies in Colom bia and immediately surrounding areas ..................................... 92 A-15 Log-likelihood map of predicted distributions for selected unidentified morphospecies in Colom bia and immediately surrounding areas ..................................... 93 A-16 Log-likelihood map of predicted distributions for selected unidentified morphospecies in Colom bia and immediately surrounding areas ..................................... 94

PAGE 10

10 A-17 Log-likelihood map of predicted distributions for selected unidentified morphospecies in Colom bia and immediately surrounding areas ..................................... 95 A-18 Log-likelihood map of predicted distributions for selected unidentified morphospecies in Colom bia and immediately surrounding areas ..................................... 96 B-1 Species accumulation curves for selected sites (1, 2) ........................................................ 98 B-2 Species accumulation curves for selected sites (3, 4) ........................................................ 99 B-3 Species accumulation curves for selected sites (5, 6) ...................................................... 100 B-4 Species accumulation curves for selected sites (7, 8) ...................................................... 101 B-5 Species accumulation curves for selected sites (12, 13) .................................................. 102 B-6 Species accumulation curves for selected sites (15, 18) .................................................. 103 B-7 Species accumulation curves for selected sites (22, 23) .................................................. 104 B-8 Species accumulation curves for selected sites (24, 25) .................................................. 105 B-9 Species accumulation curves for selected sites (26, 30) .................................................. 106

PAGE 11

11 LIST OF TERMS/SYMBOLS/ABBREVIATIONS -diversity The biodiversity or specie s richness of a single area or site. -diversity The biodiversity or species ric hness turnover between one area or site and another. Turnover is here defined as the difference between species richness between two areas or sites, as well as difference in composition of species richness between two areas or sites. -diversity The biodiversity or species ric hness of all areas and sites of the study. Biodiversity A measure of the amount of biological distinctne ss within an area and/or through a span of time. Here it is de fined as the number of species, or species richness (morphoor otherwise) in an area or through a span of time. Biodiversity and diversity are here used interchangeably. Effort The amount of work or expenditure of energy that is used to capture individuals. Here it is defi ned as the number of traps at a given site or for a given time period. EstimateS A software program that es timates diversity statistics and species accumulation curves iteratively by rea rranging sample order in specified fashions, such as with or without replacement, and pooling across iterations. These rearrangements allow for better estimates of the total diversity of the samples. Maximum Entropy The principle governing the distribution modeling software MaxEnt. Maximum entropy finds the probability distribution that is the most uniform or spread out based upon the in formation and data it is given. The probability distribution is based on incomplete knowledge of the true distribution and only presence data is required. These distributions are projected onto maps corresponding to th e layers used in the analysis. The MaxEnt software, which uses a maximum-likelihood framework, models species distributions by generati ng a probability distribution over the pixels of the map. The probability that the set of habitat conditions represented in each pixel will be suitab le for the species is shown in color format, with red indicating the stro ngest prediction and blue indicating very poor predictions (going from red > yellow > green > blue in terms of strength of prediction). PNN Parque Nacional Natural (National Natural Park) RN Reserva Nacional Natural (National Natural Reserve) SFF Santuario de Flora y Fauna (Flora and Fauna Sanctuary)

PAGE 12

12 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 FIREFLY DIVERSITY IN COLOMBIA: PATTERNS ACROSS A DYNAMIC LANDSCAPE By Bradley Williams Smith May, 2009 Chair: Marc Branham Major: Entomology The diversity of Neotropical Lampyrid ae has not been addressed since the 19th century. The charismatic nature of fireflies contributes to their use as conservation units, especially in the Neotropical region where the group exhibits much di versity. The diversity of fireflies within the country of Colombia, a designated hotspot of worldwide biodiversity, is herein quantified for first time across both spatial and temporal scal es. The population distribution modeling software MaxEnt was used to estimate the distribution of Lampyridae across Colombia as well as the distribution of individual spec ies where possible. Fishers Evenness Values, and Species Accumulation Curves were calculated via the so ftware package EstimateS for the collection sites used in this study. Firefly Abundance and Sp ecies Richness values were calculated for each month throughout the study. Differences between the geographical distributi ons of the different sexual signaling systems (pheromone, photic + pheromone, photic) were modeled using MaxEnt. Data collected over th e course of this study indicate s that firefly species are not distributed randomly across habitats in Colombia a nd that the estimated numb er of firefly species still awaiting capture within Colombia is greater than the total am ount of firefly species collected over the course of this study. In addition, each sexua l signaling system was also not distributed in a random manner. Finally, data from this study su ggests that there are di fferences between the

PAGE 13

13 geographic distributions of the each firefly sexual signaling system s. Hopefully the patterns of species distributions across time and space reported he re will be of value to future researchers in their attempts to study and conserve biodiversity in the Neotropics in general, but the country of Colombia in particular.

PAGE 14

14 CHAPTER 1 INTRODUCTION Firefly beetles of the fam ily Lampyridae are members of the Coleoptera superfamily Elateroidea (Lawrence and Newton 1995), which in cludes other bioluminescent beetles such as Pyrophorus spp. (Elateridae) and phengodid beetles (Phengodidae) (OKeefe, 2002). Fireflies are often a conspicuous and vi sible component of ecosystem s due to their production of bioluminescent emissions (Takeda et al., 2006), which the adults of some species use to advertise their gender and species (Barber, 1951; Lloyd, 1964, 1971; Mast, 1912; McDermott, 1911, 1958; Papi, 1969; Branham and Wenzel, 2003). Because of their bioluminescence, fireflies have become a part of many cultures mythologies, where they have been used to explain claims of supernatural phenomena (Kobori and Primack, 2003; Takeda et al ., 2006). Fireflies are thus valuable because of their intrinsic beauty, their cultural and historical significance, and the role they play as predators in their ecosystems. Firefly species diversity is poorly known in mo st regions of the world. This is especially true for the Neotropics, where more than half of the currently described species of fireflies are found (Lloyd, 2002). Lloyd (2002) estimated that more than 2000 species in the Neotropical region are currently undescrib ed and unknown to science. Many areas within Colombia, including the tr opical Andes and the C hoc region (tropical moist and dry forests) have been designate d as world biodiversity hotspots (Myers et al ., 2000). The diversity of species within these areas of the country as well as the large number of precinctive species (Jaramillo, 2006, Kattan et al ., 2006) has highlighted Co lombia as an area for the conservation of vertebrates and angiosperms. The paucity of data on Neotropical and speci fically Colombian fire fly diversity is the motivation for this study. The destruction of natural habitat within Colombia is rapidly

PAGE 15

15 accelerating (Etter et al ., 2006a, 2006b) and with the destructi on comes a large amount of overall species loss, including Neotropical firefly species. There is therefore a need to rapidly describe and document the distribution and divers ity of fireflies within Colombia. The objectives for this thesis are threefold: 1) analyze the geographic distribution and species richness of fireflies within Colombia and elucidate the variables most affecting their current distribution, 2) estimate the species diversity of fireflies in Colombia as well as areas within the country both temporally and spatially at both geographic and elevational scales, and 3) examine the distribution of firefly sexua l signaling modalities throughout Colombia. These three objectives attempt to answer several questions. Are fireflies randomly distributed throughout geographic and temporal sp ace? Are there environmental parameters that correlate with the current distri bution of fireflies? Do environm ental parameters differentially affect the current distribution of different sexual signaling modalitie s? With regard to previous studies on Neotropical fireflies, is Colombia more or less diverse than other countries? The distribution of fireflies within Colombia will be estimated using the MaxEnt software program, a distribution modeling program that uses the principle of maximum entropy to describe the distribution of sp ecies within a dataset (Philips et al ., 2004). MaxEnt was chosen for several reasons: it has been shown to outperform most other distribution modeling programs, it can compensate for small sample datasets and it is relatively easy to use (Elith et al ., 2006; Ficetola et al ., 2007; Gibson et al ., 2007; Guisan et al ., 2007; Hernandez et al ., 2006; Papes and Gaubert, 2007). The distribution of firefly species diversity across Colombia (21 areas and parks, 4205 traps over 4 years) will be modeled, as well as each distinct sexual signaling modality. This will be the first time that the sexual signaling m odalities of fireflies have been quantified and modeled for any geographic region.

PAGE 16

16 The distribution of firefly species diversity will be estimated usi ng morphospecies as a surrogate for species (Rees, 1983; Oliver a nd Beattie, 1993, 1996a, 1996b; Beattie and Oliver, 1995; Pik et al 1999; Derraik et al ., 2002) and the diversity es timation software program EstimateS (Colwell, 2005). The taxonomy of Colombian firef lies is poorly known and it would be extremely difficult and very time consuming fo r a correct genus and species determination to be made for the majority of the species that have been described. There are no known living entomologists that specialize in the taxonomy and identification of this fauna, and type specimens are scattered among many collections in both South America and Europe. The collections of deposition for many important type specimens are unknown. The EstimateS algorithm was used to estimate a suite of diversity statistics (such as Chao 1 & 2, ICE, ACE, Simpson and others) for this fauna by randomly choosing a sample from the sample space and recording the species in it and the abundance of each. I here attempt to describe the actual distribution of the Colomb ian firefly fauna (in terms of both abundance and species richness) across elev ation and time as well as their potential distribution given a number of diffe rent environmental variables. I also partition the data into bins according to the sexual signaling systems used by this fauna to invest igate whether there are additional distribution patterns acro ss the different types of firefly signal systems. My working hypothesis is that fireflies, bot h in terms of species richness and abundance, are not randomly distributed across space and time.

PAGE 17

17 CHAPTER 2 LITERATURE REVIEW Fireflies are a charism atic group of beetles belonging to the family Lampyridae. The beetles are well known to the general public and have several colloquial names including fireflies, glowworms and lightning bugs, alt hough glowworm also includes phengodid beetles (larvae and adult females) and the larvae of some mycetophilid flies (Lloyd, 2002). These beetles have in the past been mythologized as fairy lights and other supernat ural phenomena (Takeda et al ., 2006). Fireflies occur on all continents except Antarctica. The last catalog of the family (McDermott, 1966) listed 1891 species in seve n subfamilies and 92 genera. A more recent estimate by Lloyd (2002) placed th e number of described spec ies at approximately 2000. The Neotropical firefly fauna contains more than half of the described species (Lloyd, 2002). This fauna is poorly known however, and it is esti mated that more than 2000 species remains undescribed in the Neotropics. Firefly biology is known only tangentially for mo st species, as relativ ely few complete life history studies have been conducted (McDermott, 1964). The larvae of all known fireflies are luminescent to some degree (Sivinski, 1981; Labella and Lloyd, 1991; Archangelsky and Branham, 2001; Branham and Wenz el, 2001) and the glow is emitted in almost all cases from a pair of simple lanterns on abdominal segment VIII (Lloyd, 2002). Luminescence is hypothesized to have been first used by firefly larvae as an aposematic signal and then co-opted in the adult stage for sexual communication (Branham a nd Wenzel, 2001, 2003). Larvae can be terrestrial, semi-aquatic or aquatic depending upon the spec ies and are found in a variety of habitats including old fields, rotten logs and around ponds and marshes (Lloyd, 2002). All larvae are predaceous. Pupation often occurs underground and can require up to eight months for completion (McDermott, 1964).

PAGE 18

18 Although all known larvae of fireflies glow not all adults do (Branham and Wenzel, 2003; de Cock and Matthysen, 2005). Adult fireflie s are often a conspicuous component of their ecosystem, especially when the species emit n eurologically precise flashes and pulses (Lloyd, 2002). These flashes are what consti tute the charismatic nature of fireflies. The bioluminescent emissions of fireflies are used to advertis e sex, species and locati on (Barber, 1951; Lloyd, 1964, 1971; Mast, 1912; McDermott, 1911, 1958; Papi, 1969; Branham and Wenzel, 2003). Lloyd (1968) hypothesized that the flas hing may also have non-sexual functions, including illuminating landing sites and incorporation in femme-fatale behavior (Lloyd, 1965). The luminosity of adult fireflies, in addition to its se xual function, also appears to be a type of warning coloration (Cowles, 1959; McDermott, 1964) to advertise unpalatability or a poisonous nature (Marc Branham, personal observation). There is a cons iderable amount of variation between flash patterns among fireflies (Lloyd, 2002), ranging from long drawn out emissions to short flashes to flash-pulse schemes. Determining the species identity within some genera, such as Photuris can be difficult, if not impossible, if no informati on on the flash pattern is associated with the specimen (Lloyd, personal communication). Rather th an the sole use of bioluminescent signals, courtship signaling in adult fireflies can take the form of a pheromonal signal system (without bioluminescence) as well as a combination of pheromones + photic emission to attract or locate mates (Gorham, 1880; McDermott, 1911; Mast 1912; McDermott, 1914; Hess, 1922; Blair, 1924; Ohba, 1983; David and Birch, 1989; Branham and Wenzel, 2003). Colombia, South America, is a megadivers e country in which many different taxonomic groups have a high degree of precinctivity (Myers et al ., 2000; Wyngaarden and FandioLozano, 2005; Jaramillo, 2006; Kattan et al ., 2006). It is hypothesized th at the same is true for fireflies. The geography of the country is quite varied, ranging from coastal plains and mangrove

PAGE 19

19 forests to glacial grasslands, w ith the elevation ranging from sea level to more than 4000 meters (Wyngaarden and Fandio-Lozano, 2005). Many habitats within Colombia are highly stressed by human disturbances (Etter et al ., 2006a, 2006b; McNeely 2003) and its natural areas are rapidly diminishing. Because of this, firefly biodiversity is being lost at the same rate that their habitat is being destroyed or irrepa rably altered (Lloyd, 2002). Biological diversity, or biodi versity, can be quantified at three different levels: and (Loreau, 2000). The diversity refers to the total numbe r of species occurring within a unit area. The overlap in species composition between different units of area is the diversity. The overall diversity for all of the ecosystems or regions being examined is the diversity. Local and regional biogeographical processes work sy nergistically to govern biodiversity (Loreau, 2000). Delong (1996) gave a synthesi zed definition for biodiversity: Biodiversity is an attribute of a site or area that consists of the variety within and among biotic communities, whether influenced by humans or not, at any spatial scale from microsites and habitat patches to the entire biosphere. The biodiversity of an area can be defined on many different levels, fr om the diversity of domains and phyla within a country down to the number of species and the degree of genetic variation within species found in a leaf litter sample. The definition of biodiversity that is used for a study must be carefully described a nd rationalized for the purpose of the study. Because of the overwhelming diversity and numbe r of arthropod species, it is often very difficult to place the correct Latin name w ith a specimen (Oliver and Beattie, 1993, 1996a, 1996b). Species determinations are not always a straightforward endeavor, as the taxa being identified may have a convoluted literature surrounding it (Oliver and Beattie, 1993). In addition, if it is a poorly studied group then a formal desi gnation may not be possible (Oliver and Beattie, 1993). The use of morphospecies is often the only av ailable tool for research ers trying to assess the biological richness of habita ts as well as for researchers who are studying invertebrates

PAGE 20

20 (Derraik et al ., 2002). There are too few specialized invert ebrate taxonomists for every project to make use of them (Derraik et al ., 2002; Oliver and Bea ttie, 1993), especially if groups such as Coleoptera are being examined (Derraik et al 2002, Oliver and B eattie, 1993, 1996a, 1996b). Funding and time issues often limit what taxa can be identified to the species level in biotic inventories, and these are most often vertebrates and flowering plants (Oliver and Beattie, 1993; Oliver and Beattie, 1996a; Palmer 1990, 1991). Invert ebrates comprise much of the diversity in most ecosystems, and because they are often ab undant (both in numbers of individuals and species) they should be just as important to inventories as vert ebrates and angiosperms (Oliver and Beattie, 1993). However, usef ul surrogates for invertebrate biodiversity are few and far between. It has been found in some groups (Briti sh butterflies and dragonflies (Prendergast et al ., 1993)) that areas of high diversity for one group do not necessarily correlate with areas of high diversity in another group (Oliver and Beattie, 1996a). In these instances where disparate groups do not correspond, a shopping basket approach may be effective (Oliver and Beattie, 1996a). Using this approach, taxa should be identified th at have diversities which correlate with one another, or which demonstrate responses whic h are similar with respect to environmental gradients or disturbance. To save time and resources, a method of desi gnating discrete biologi cal units other than species has been recommended (Oliver and Beattie, 1993, 1996a, 1996b; Derraik et al 2002; Pik et al ., 1999). Oliver and B eattie (1993) have called this unit the Recognizable Taxonomic Unit (RTU), or morphospecies. They re port that the term was first used by Rees (1983) to aid in sorting Sulawesian Hemiptera. The definition given by Rees (1983) was [RTUs are] taxa that differ from each other clearly. A clear differe nce referred to morphologi cal variation in the specimens; he acknowledged that an RTU was not necessarily a good (read: taxonomic)

PAGE 21

21 species. Although the term RTU is still in use, many use the equivalent term morphospecies (Oliver and Beattie, 1993, 1996a, 1996b; Derraik et al ., 2002; Pik et al 1999; Beattie and Oliver, 1995). In their findings, Oliver and Beattie (1996b) found that mo rphospecies obtained similar results when compared to taxonomically define d species. Oliver and Beattie (1993) also found that splitting (defined as a single taxonomi c species being split among more than one morphospecies) and lumping (defined as multiple taxonomic species being lumped into one morphospecies) effects were able to counterbala nce each other and reduce the overall error. They found that it was important to know how the inclusion of different groups of taxa will affect the error rate, and that it wa s probably best to use as few taxa (and life stages) as possible to minimize the adverse effects. Krell (2004), however, found that of 79 studies using morphospecies, only three estimated the correct number of species. The other studies were fairly evenly distributed between overand underestimations (41 and 35 studies, resp ectively). Thus, it can be seen that sorting to morphospecies yields an approximately even distri bution of conservative a nd liberal estimates of species number. He stated that, The low e rror rates were only good luck, caused by a similar number of parataxonomic splittings and lumpings. and, It may be seriously questioned if a high level of inaccuracy in a sorti ng result is acceptable if the gross error is low, because the low overall error is caused only by good luck. Just having a lot of data is not enough for something to be put forth as valid, exact and reliable, even if the statistics back it up (Krell, 2004). For ecological and biodiversity stud ies, morphospecies voucher sp ecimens are not (necessarily) equivalent to taxonomic voucher specimens (Brower, 1995; Krell 2004). Morphospecies by definition are not taxonomically defined, and thus if more than one speci men of a morphospecies series exists, there could potentially be mu ltiple species represented by several vouchers

PAGE 22

22 (Brower, 1995; Krell, 2004). Ca mpbell (1995) stated that, an alternative explanation of the data is that the morphospecies approach appear s to be successful wher e the taxonomy is poorly known and keys to spec ies are inadequate. Studies which use the morphospecies concept are not concerned with formal description or naming of the species they enc ounter (Oliver and Beatti e, 1993). It is used as a time saver both to sort the specimens to original morphospecies, as well as aiding the prof essional taxonomist in more quickly identifying taxonomic species (Oliver and Beattie, 1993, 1996a, 1996b; Derraik et al ., 2002; Pik et al ., 1999). Despite his obvious dislike of the morphospecies concept, Krell (2004) believed that morphospecies (where appropriate) should be used as an instrument to heuristically find meani ngful patterns when the taxonomic desi gnation of specimens is either too difficult or impossible. Eliminating the morphospeci es concept outright and in all applications is not necessary and would hinder otherwise daunting and useful studies (Krell, 2004). The question, How many species are there? is usually not as straightforward as one might expect that is, the procedure of simply count ing the number of species (Bunge and Fitzpatrick, 1993; Chazdon et al ., 1998; Colwell and Coddington, 1994; Go telli and Colwell, 2001; Walther and Moore, 2005; Hortal et al ., 2006). For organisms such as invertebrates, it is almost impossible to sample all of the diversity for vi rtually any taxon in a na tural environment. The effects of the sampling regime on the collection of any one sample can bias what organisms are caught within it (Leather and Watt, 2005). Standardizing the sampli ng effort can help in allowing the results of a sampling regime to be compared across studies. It should be noted however, that sampling regimes from one area applied to a new area will not necessarily produce similar, or even comparable, results (Hortal et al ., 2006).

PAGE 23

23 Because sampling rarely encompasses the en tire diversity of any taxon, extrapolative methods have been applied in an attempt to estimate the true number of species in a given area (Colwell et al ., 2004a; Gotelli and Colwell, 2001). Species accumulation curves are a method of plotting the number of species co llected in a sample and iteratively pooling all the new species for all samples (Soberon and Llorente, 1993; Ugland et al ., 2003). This can be done in several ways, such as successively pooling all samples by sample number or by successively pooling all samples by the period in time in which they were set. By using statisti cal software, such as EstimateS, diversity statistics, such as species richness estimators, can be calculated (Colwell, 2005). The highly stressed ecosystems of Colombia are hypothesized to represent a large amount of firefly diversity, both precinctiv e and otherwise. Because of this, the estimated and potential distribution of fireflies within the country needs to be quant ified and mapped. Describing the distribution of fireflies across Colo mbia requires the use of models to assess what parameters are important to the distribution of these beetles. Wolda et al ., (1998) found that species diversity and abundance in other Neotropica l taxa can be greatly affected by the seasonality of the area studied. Other physical aspects of an organisms environment, such as elevation, annual precipitation and temperature, can also play a role in the realized distribution of the organism (Wolda et al ., 1998). The distribution of species acr oss geographical ranges, much like the question of, How many species? would seem to be able to be answ ered in a straightforward fashion: plot the occurrence records on a map and the distribution is revealed. Again, much like the question of species number, the question of where species ha ve been found to occur and where they do occur do not always produce the same answer. Another important consideration when examining

PAGE 24

24 species distributions is th e grain size that one will be looki ng at, or the size of the individual units being considered (Guisan et al ., 2007). Phillips et al (2006) have created a program called MaxEnt, which uses the principl e of maximum entropy, to extrapol ate species distributions from locality information. The principle of maximu m entropy states that the distribution which maximizes the amount of entropy within a system is the best one. Entropy can be equated to the distribution which is the most spread out while at the same time avoiding the placement of unfounded constraints (Pearson et al ., 2007). Gibson et al (2007) stated that, In general, MaxEnt is a method for making inferences from in complete information and in this case [their study], approximating an unknown pr obability distribution. Several software packages have been designe d to model the distribution of organisms. MaxEnt (Philips et al ., 2004) uses the principal of maximu m entropy to model the distribution of species. Given the constraints of the vari ables input into the program and the species information uploaded, MaxEnt outputs a probabili ty distribution that maximizes the amount of entropy in the information, or creates the maxi mally uniform distribution for a given taxa (Phillips et al ., 2004, 2006; Gibson et al ., 2007). Only presence data is needed, as MaxEnt creates pseudo-random background points with which to check for statistical significance. MaxEnt has been shown to consistently outpe rform almost all other distribution modeling programs with respect to both model performan ce parameters (AUC [area under the curve] and ROC [receiver operating characte ristic] curves) (Ficetola et al ., 2007; Gibson et al ., 2007; Guisan et al ., 2007; Papes and Gaubert, 2007; Elith et al ., 2006) as well performance when small numbers of presence points are used (Hernandez et al ., 2006). Hernandez et al (2006) stated that, MaxEnt can somewhat compensate for inco mplete, small species occurrence data sets and perform near maximal accuracy in this [sic] conditions. MaxEnt has been shown to out

PAGE 25

25 perform GARP which is a similar, though older, species distribution modeling algorithm (Philips et al ., 2006).

PAGE 26

26 CHAPTER 3 FIREFLY DISTRIBUTIONS AND DIVERSITY AC ROSS THE GEOGRAPHIC RANGE OF COLOMBIA Introduction The Neotropical region is am ong the most species rich areas of the world (Erwin, 1988). Within this region, Colombia has been designated as one of the top biodiversity hotspots of the world (Myers et al ., 2000). Much of the vertebrate fauna of Colombia, especially birds, has been well characterized (Jaramillo, 2006; Kattan, 1992; Kattan et al ., 2006). However, the insect fauna within Colombia remains poorly character ized (Gaston, 1991), and if the situation for Colombia is similar to that of its geographic neighbor, Panama, then there is likely to be many fold more undescribed species awaiting capture than there are described species (Wolda et al ., 1998). Neotropical fireflies were last treated as a whole in McDe rmotts (1966) review of the world firefly fauna. Based on the relatively few described species known to occur in Colombia, the diversity of the Colombian firefly fauna clearly appears to be poorly known. Lloyd (2002) has estimated that the number of described fireflies worldwide (approx imately 2000) is two to three fold times lower than the number of unde scribed species. Further, Lloyd (2002) estimated that more than half of the already described species are Neotr opical. Therefore, it is readily apparent that the Neotropical firefly fauna is poorly known, with most of its diversity being unknown to science. White (1975) estimated that the total species richness of Cantharidae, Lycidae and Lampyridae in the United States was 750 species for all three of these families. His method of estimation involved plotting the num ber of valid species recorded by decade and drawing a free hand curve that would join the maximum number of points while leaving an approximately even distribution of points to the right and left of each line so that th e upper half of the curve would be

PAGE 27

27 an approximate mirror image of the lower half Whites (1975) method was necessarily less complex than recent species estimates as the co mputing power, especially of computers owned by individuals, was much less than it is today. However, his estimation of cantharoid diversity for a large area was one of the first attempted. The geography and physiography of Colombia is extremely diverse, with an estimated 337 ecosystems and 63 chorological types (Wyngaarde n and Fandio-Lozano, 2005), and elevations ranging from sea level to more than 4000 meters above sea level. A chorological type according to Wyngaarden and Fandio-Lozano (2005) consis ts of groups of spatial [sic] related ecosystems that account for biodivers ity process [sic] that operate at larger scales, as well as being a group of spatially related ecosystems that are clustered toge ther such that they are more similar to each other than to any other group. Th is wide range of geograp hic features helps to account for the great diversity of life within Colombia. The distribution of firefly species within Colombia was modeled using the MaxEnt computer program, which does not require absen ce data. The diversity of Colombian fireflies was assessed using the program EstimateS. A wide scale study of firefly distributions and diversity to date has not been car ried out, and the results of this study could help to guide further research on fireflies both within Colombia as well as setting a foundation for future studies to expand upon and refine. Perhaps most importantly, quantifying the firefly fauna of Colombia will serve as an important base-line for future research on the biological diversity in Colombia and the conservation of this natural resource. Materials and Methods Specimens The specim ens used for this study were collecte d as part of the NSF funded project Insect Survey of a Megadiverse Country, Phase II: Colombia, Dr. Michael Sharkey PI, from January,

PAGE 28

28 2000 to December, 2003 across the country of Colombia. A total of 4205 Malaise traps were placed in 25 National Parks or similarly protected areas and were maintained for the four year period of the study (Table 3-1 and Figure 3-1). On average 87.6 traps were set during each month. While the Sharkey Lab was primarily in terested in documenti ng the biodiversity of parasitic hymenoptera in Colombia, a number of firefly and phengodid beetle s were incidentally collected in the Malaise traps. All resulting specimens were s ubsequently sent to Dr. Marc Branham for analysis. These specimens were then pinned and labeled and sorted to morphospecies to later be identified as accurately as possible. Distribution Modeling MaxEnt (ver. 3.1.0) (2007) was used to m odel the distribution of species across Colombia. MaxEnt is a softwa re program that utilizes th e principle of maximum entropy (Philips et al ., 2004; Philips et al ., 2006), meaning that the distribu tion of an object is most likely to be the one which is the most spread out (maximized entropy), or more formally defined, it is the distribution which agrees with all of the environmental conditions at the occurrence points but that avoids placing unfounded constraints on the data (Pearson et al ., 2007). This modeling software does not require presence/absence data for the collection points, only presence. Stockwell and Peterson (2002), us ing the program GARP (Genetic Algorithm for Rule-set Prediction), found that using presence only records did not present a difficulty for accurate distribution predictions. The MaxEnt program will choose an amount (specified by the user) of pseudoabsence pixel po ints randomly on the map and use these in lieu of absence points. These pseudoabsence points allow MaxEnt to test for the significance of the models predictive power without having to use true absence points; wit hout pseudoabsence points there would be nothing for the model to compare presence data to. This aspect of MaxEnt is one of

PAGE 29

2 9 Table 3-1. Sites within Colombia represented by collected firefly specimens Site Site # Area (km2) Range of altitude (m above sea level) SFF1 Iguaque (Boyac) 1 67.5 2400 3800 PNN Gorgona (Cauca) 2 616.875 0 -300 PNN2 Amacayacu (Amazonas) 3 2935 100 180 PNN Tayrona (Magdalena) 4 150 0 900 RN3 La Planada (Nario) 5 32 1850 1930 PNN Sierra Nevada de Santa Marta (Magdalena) 6 3830 0 5775 PNN Serrana de Chiribiquete (Caquet) 7 12800 200 1000 PNN El Tuparro (Vichada) 8 5480 0330 PNN Utra (Choc) 12 543 0 1200 PNN Chingaza (Cundinamarca) 13 533.85 800 4020 SFF Los Colorados (Bolivar) 15 10 180 400 PNN Cordillera de Los Picachos (Caquet) 16 4470 450 3500 PNN Farallones de Cali (Valle del Cauca) 18 1500 200 4100 PNN Sierra de La Macarena (Meta) 22 6292.8 400 2500 PNN Cueva de los Gucharos (Huila) 23 90 450 Data not available PNN La Paya (Putumayo) 24 4220 200 330 PNN Sumapaz (Cundinamarca) 25 2935 Data not available Estacin Biolgica Mosiro-Itaj ura (Capar) (Vaups) 26 Data not available Data not available SFF Otn Quimbaya (Risaralda) 27 4.89 1800 2900 Leticia Va Tarapac (Amazonas) 28 Data not available Data not available PNN Los Katos (Choc) 30 720 2 30 1) Santuario de Flora y Fauna (Flora and Fauna Sanctuary) 2) Parque Nacional Natural (National Natural Park) 3) Reserva Nacional Natural (National Natural Reserve)

PAGE 30

30 Figure 3-1. Map of Colombia showing all collecti on sites (not all sites yi elded fireflies). Site numbers correspond to those in Table 3-1. Obtained from http://www.sharkeylab.org/biodivers ity/db.php?app=colombia&function=sites on June 24, 2008

PAGE 31

31 its most desirable, as presence/absence data is difficult to obtain for many species, especially those that are rare, difficult to co llect or detect, or do not utiliz e the entire environment that is suitable for supporting them (Elith et al ., 2006; Gibson et al ., 2007). MaxEnt has been shown to outperform other distribution modeling software programs, especially when sample sizes are low (Hernandez et al ., 2006; Gibson et al ., 2007; Guisan et al ., 2007; Papes and Gaubert, 2007). Hernandez et al (2006) found that MaxEnt sample sizes of 50 allowed the AUC (Area Under Curve) to reach near maximal levels and that the predicted area remained relatively stable with 25 samples for MaxEnt. The AUC describes the amount of area underneath the Receiver Operating Characteri stics (ROC) curve, whic h is itself a ratio measure of the number of true positive results compared to false positive results (Metz, 1978; van Erkel and Pattynama, 1998). An area of 1.0 corre sponds to a model that perfectly predicts matches the data (all true positives), while an area 0.50 corresponds to a model that matches the data only 50% of the time (equal number of true and false positives). The AUC is used as a surrogate for accuracy of the model in MaxEnt. Hernandez et al (2006) also found that useful models were created with as few as 5-10 presence observations and that as few as 50 presence observations produced models that were essentially the same as those with twice as many presence observations. This small-model competency is thought to be due to MaxEnts regularization procedur e which counteracts the problem of overfitting (Hernandez et al ., 2006). The regularization pr ocedure causes the average value for a predictor to be close to, but not e qual to, the empirical average. This effectively smoothes and relaxes the fitting process of the algorithm, compensating for MaxEnts predilection to over fit models (Phillips et al ., 2004; Hernandez et al ., 2006).

PAGE 32

32 The spatial distribution of a species (or group of species) is ca lculated by inputting presence or collection locality in formation and then correlating these points to the environmental data layers being used in th e analysis. An environmental data layer is a mathematical representation of the geographical area being an alyzed. Examples of environmental layers include: average rainfall, elev ation, average temperature, etc MaxEnt begins with a perfectly uniform probability distribution of the species and then iteratively alters the weights of the data layers in order to maximize the likelihood of the presence points in the occurrence dataset (Hernandez et al ., 2006). To test the effectiveness of the generated model, MaxEnt uses training and testing points. MaxEnt is trained using a certain user-defined percentage of the data to create a model based upon the presence localities and the data layers us ed. The testing points are run using the created model and how well they correspond to it is give n by several different parameters, such as the Receiver Operating Characteris tic (ROC) and the (AUC). For this study, 20 bioclimatic and elevation layers from the WorldCLIM dataset were downloaded (http://www.worldclim.org/) (Hijmans et al ., 2005) and used to model the distribution of fireflies across Colombia (Table 3-2). These laye rs consisted of generic 30 arcsecond grids of current (as of 2005) conditions in two separate datasets that included Colombia (tiles 23 and 33 on WorldCLIM website). These da tasets were stitched together (merged) following the instructions on the website in the co mmand line program Arc. In order to be usable in MaxEnt the layers were then converte d to ASCII files in Ar cGIS ArcMap 9.1 (ESRI Corporation, Redlands, CA). The MaxEnt program was set to run for a maximum of 10,000 iterations with 30% of the samples being set aside for model testing. All firefly morphospecies were combined into a single

PAGE 33

33 Table 3-2. WorldCLIM Data layers used in analysis Layer Data being displayed ALT Altitude BIO1 Annual Mean Temperature BIO2 Mean Diurnal Range (Mean of monthly (max temp min temp)) BIO3 Isothermality (P2/P7)(*100) BIO4 Temperature Seasonality (standard deviation 100) BIO5 Max Temperature of Warmest Month BIO6 Min Temperature of Coldest Month BIO7 Temperature Annual Range (P5 P6) BIO8 Mean Temperature of Wettest Quarter BIO9 Mean Temperature of Driest Quarter BIO10 Mean Temperature of Warmest Quarter BIO11 Mean Temperature of Coldest Quarter BIO12 Annual Precipitation BIO13 Precipitation of Wettest Month BIO14 Precipitation of Driest Month BIO15 Precipitation Seasonality (Coefficient of Variation) BIO16 Precipitation of Wettest Quarter BIO17 Precipitation of Driest Quarter BIO18 Precipitation of Warmest Quarter BIO19 Precipitation of Coldest Quarter P2, P5, P6, P7 refer to precipitation layers not used in this analysis. species such that MaxEnt would effectivel y model the distribution of the firefly family, Lampyridae, across Colombia. The Jackknifing procedure was selected to determine the contribution of each layer vari able in the analysis. The Random Seed setting was selected to alter the pixels used for the pseudoabsence points during each iteration and the No Duplicate presence localities setting was used for the analysis Elith et al (2006) found that collection biases for studies involving distribution mode ling can have drastic effects. Although more than 500 traps caught fireflies, only 52 unique lo cations yielded firefly specimens. Many of the traps were either clustered together in a small, ecol ogically homogenous area an d/or traps were set sequentially over a period of time in the same lo cation. These clusters and sequences of traps had identical latitude and longitude coordinates and many did not ha ve a unique geographic identity. All other parameters were unchanged from defau lt settings. The threshold criterion of maximum

PAGE 34

34 test sensitivity plus specificity was chosen to validate the model, and test Area Under Curve (AUC) was used as a surrogate to determine how closely the test data matched the training data. Estimating Diversity The EstimateS (Colwell, 2005) (ver. 8.00) so ftware program was used to estimate the amount of species diversity bot h throughout Colombia and within the individual trapping sites (Table 3-1). EstimateS estimates diversity stat istics by randomly choosi ng a sample out of the sample space and recording what species (and the abundance of each species) was found in that sample. Depending upon the options chosen, this can be done either with or without replacement of that sample back into the sample space. Sa mpling with replacement means that a sample could potentially be chosen and entered into the calcul ation more than once during a run given that all samples have equal probability of being chose n. Sampling without replace ment means that after a sample is chosen and added to the dataset it is removed and cannot be chosen again. Thus, sampling without replacement means that all sa mples will always be used for each run, while sampling without replacement means that often not all samples will be used for each run. This randomness of sample addition is, along with values being averaged over successive runs, the most important aspect of EstimateS. By ra ndomly choosing samples, it allows the diversity statistics to be representative of the samples and not upon the sample order. A wide variety of diversity statistics, such as the Simpson index, Incidence-based coverage estimator (ICE) and abundance-ba sed coverage estimator (ACE) are computed and the average for each statistic is retained for each sampling of the dataset (Colwell, 2005 for an exhaustive listing). Thus, if 20 samples were run 100 times, then the average of the first sample of the dataset would be recorded and kept in memor y, as are the averages for the other 19 dataset samples. This retention of averages allows for extrapolation of Species Accumulation Curves for

PAGE 35

35 the diversity statistics (Colwell et al., 2004). The slope of the accumulation curve represents the rate of new species being added and can indicate the degree to which species diversity has been exhaustively sampled. Species accumulation curves can be one of two types: observed or estimated (Sobern and Lorente, 1993). Observed species accumulati on curves show how ma ny actually collected species are being added to the pool. Estimated species accumulation curves extrapolate beyond the addition of more collected species to how many species, gi ven the number collected, are estimated to truly be present (Col well and Coddington, 1994; Magurran, 2004). For the datasets being analy zed here (samples pooled from all sites across Colombia, samples from each separate site pooled separate ly), EstimateS was run for 1000 iterations of each dataset. The matrix for each dataset was setup by having samples as rows and species as columns. The dataset was randomized without replacement (Toti et al ., 2000; Thompson et al ., 2007) and Fishers and Evenness of Diversity values we re computed. All other parameters were unchanged. Fishers was calculated by the following two equations: 1st:x xxNS 1ln/1/, where S = the number of species, N = the total number of individuals. The x -value is then iteratively so lved for the ratio of S/N. 2nd: x xN 1, where N = the number of individuals and x = the value from the previous equation (Fisher et al ., 1943; Magurran, 2004). The Evenness value was cal culated by the following two equations: 1st: 2ipD, where p = proportion of total individuals in the i th species (Magurran, 2004). 2nd: S D E /1 where D = the value from the previous equation and S = the total number of species (Simpson, 1949). There were too few samples in SFF Otn Quimbaya and PNN Sumapaz for EstimateS to calculate diversity statistics.

PAGE 36

36 To estimate the species accumulation curves the Incidence-based Coverage Estimator (ICE) was used (Lee and Chao, 1994). The ICE statis tic is complicated to compute. In order to compute Sice, it is important to note a proof. It should be shown thatfreq r obsSSS inf, where Sobs = the total number of observed species, Sinfr = the total number of in frequently encountered species (found in ten or less samples), and Sfreq = the total number of frequently encountered species (found in more than ten samples). In order to calculate ICE, the C statistic must first be calculated: r iceN Q Cinf 11 where10 1 inf j j rjQ N, where j = number of samples, Qj = number of unique species (singletons) that occur in j samples. The Sice statistic is then calculated through the following equation: 2 1 inf ice ice ice r freq iceC Q C S SS, where 0,1 )( )1( )( max2 inf 10 1 1inf inf inf 2 r j j r r ice r iceN Qjj m m C S where minfr = number of total samples that have at least one infrequent species (Col well, 2005). Coverage based estimat ors, like ICE, recognize that species that are widespread or found in abundance are more likely to be included in samples than those that have restricted di stributions and low numbers of individuals (Magurran, 2004, Chao et al ., 2000). Thus, the widespread and abundant speci es do not add as much information as do those with restricted or low a bundances. Previous studies (Chazdon et al ., 1998; ZaragozaCaballero et al ., 2003; but also Williams et al ., 2007) have shown that ICE was a good estimator of species richness based on three fe atures of an ideal species richne ss estimator: 1) that it be independent of sample size beyond a threshold, 2) it would be insensitive to the patchiness of the species distributions and 3) it would be not be affected by sa mple order (Chazdon et al., 1998).

PAGE 37

37 Zaragoza-Caballero et al (2003) used ICE in their study of Mexican cantharoids to estimate species richness. Results Distribution Models A log-likelihood m ap of predicte d species distributions was cr eated from the MaxEnt run (Figure 3-2). The maximum test sensitivity plus specificity logistic threshold was 0.053 (p = 7.672E-6) and this area included 36.6% of the map. This threshold value i ndicates that a highly significant portion of the training sites are included in this area. The test omission rate was 0.067. The test AUC was 0.829 out of a possible 0.917, which indicates a strong prediction. The jackknifing procedure found that BIO19, BIO17 and A LT (Table 3-2) contribu ted the most to the MaxEnt model (58.9%, 9.5% and 9.2%, respectively). Log-likelihood maps for individual morphospecies, when able to be ge nerated, can be found in Appendix A. Diversity Estimation The species richness for Colom bia as a whole was 273 morphospecies. The species richness values ranged from one morphospecies collected (PNN Cordillera de Los Picachos, SFF Otn Quimbaya and Leticia Va Tarapac) up to 69 species caught (PNN Amacayacu). The abundance of individuals collected for Colomb ia as a whole was 2029 individuals. Abundances of individuals collected ranged from one individual (PNN Cord illera de Los Picachos and SFF Otn Quimbaya) to 700 individuals (PNN Farallones de Cali) (Figure 3-3 and Appendix B). Diversity and Evenness values va ried across parks. Fishers index of diversity ranged from a low of 0.53 (Leticia Va Tarapac) to a high of 76.04 (PNN Cueva de los Gucharos) (Figure 3-4). The Evenness ranged from zero (Le ticia Va Tarapac) to 0.9591 (PNN Cueva de los Gucharos). For all of Colombia, the Fishers index of diversity was 84.9 while the Evenness measure was 0.7345.

PAGE 38

38 Figure 3-2. Log-likelihood map of predicted firefly distribution for Colombia and immediately surrounding areas. Warm colors (i.e. red) indicate a higher like lihood of occurrence. White squares indicate training points, purple squares indicate test points

PAGE 39

39 The species accumulation curves for the country of Colombia are presented graphically in Figure 3-4. The ICE curve estimated 673.45 specie s are yet to be found in the country of Colombia; the observed species richness was 273. After 543 sample s the ICE estimator had still not begun to asymptote. The observed species accu mulation curve, while still rising, has a much lower slope than does the ICE estimator. The species accumulation curves for all individual parks can be found in Appendix A. Discussion The MaxEnt distribution m ap showed that fireflies are not random ly distributed across the geographic range of Colombia (Figure 3-1). Th e log-likelihood of a firefly being in a specific area is highest (p 1) near the northwest coast of Colo mbia and along portions of the Andean range; these areas include PNN Farallones de Ca li, PNN Sumapaz, and PNN Utria. In their study, Wyngaarden and Fandio-Lozano (2005) labeled these areas as floodplains with riparian forests, denudational slopes with semi-deciduous forests and denudational slopes with evergreen forests. A large area encompassing mainly tropical lowland forests south of the Andes also has a high likelihood (p .5 .8) of firefly species being dist ributed there. These areas correspond to PNN La Paya and PNN Chiribiquete. Wyngaar den and Fandio-Lozano (2005) labeled these areas as sedimentary plains with evergreen forests, structural hills with evergreen forests and dissected sedimentary plains with savanna gr asslands and wooded grasslands. Few of these areas, however, contain national pa rks, sanctuaries or reserves. The three most important layers contributed to more than 75% of th e variation within the dataset. These layers were precipitation of the coldest quarter, precipitatio n of the driest quarter, and altitude. Therefore, the distribution of firef lies, at least in Colombia, appears to be most affected by levels of precipita tion during dry, cold conditions as well as altitude. The altitude of

PAGE 40

40 Figure 3-3. Morphospecies and individuals caught across parks

PAGE 41

41 Figure 3-4. Fishers values and Evenness across parks. Fishers computes the diversity of a sample by multiplying the total number of individuals by the iterativ ely calculated species/indivi dual ratio. Evenness analyzes the degree to which a sample is even. A sample that has an even distribution of individuals (more equal number of species to individuals) is given a higher value than one that has a skew ed distribution of indi viduals to species.

PAGE 42

42 Figure 3-5. Pooled species accumulati on curves. Solid line is observed (Sobs Mao Tau) species richness; dashed line is estimated (ICE) species richness.

PAGE 43

43 the areas with the highest log-likelihood of fire flies being distributed th ere were areas along the coast near sea level a nd within the Andes, from 1000 2200m Perhaps fireflies in Colombia have adapted, not only to harsh weather and seasona l patterns, but also to extremes in elevation. However, a much larger portion of the map wa s covered by areas of middle elevations. This greater amount of area could offset the extremes witnessed in the highe st log-likelihood areas, meaning that more widespread species could inha bit many different types of habitat. Elith (2006) showed that organisms that are considered to be specialists generally had higher AUC values than generalists. This may be because specia lists, by their very nature, have a more limited distribution. The AUC values refer to how closely the tested site s matched the prediction of the training model, and these higher AUC values show th at it is easier to mo del specialist organisms than generalists. Diversity and Evenness values did not corr espond with abundance of individuals and species. The park where the most individuals (700) and the second most number of species (60) were collected, PNN Farallones de Cali, had a low Fishers value (15.75) and a low evenness value (0.6033) when compared to PNN Cueva de los Gucharos, a park with much lower individual (13) and species (12) abundances, whose Fishers value and Evenness values are 78 and 0.9591 respectively. This disparity is due to PNN Farallones de Cali having a large amount of a single species making up nearly half of indivi duals caught in this park. Similar disparities in diversity, seen in PNN Amacayacu and PNN Gorgona, can be accounted for in this same way. Several morphospecies were caught at two or more sites across the study area (Appendix C). However, since many of the morphospecies we re represented by a single specimen many are represented in only one area. Tent atively, these morphospecies could be said to be precinctive to these areas. More extensive trapping would need to be done, however, to be able to say

PAGE 44

44 definitively whether or not these morphospecies are found only in these areas and not other, including areas outside of Colombia. The species accumulation curves (both estimated and observed) show interesting results. Lloyd (2002) estimated that two or three times the amount of described firefly species (~2000) still remain to be discovered. The South Americ an firefly fauna is es pecially diverse (Lloyd, 2002), and the estimated species accumulation cu rve produced here supports this prediction. Both the observed and estimated curves, after 543 samples, have not begun to asymptote, although the observed curve is much less steep than the estimated (Figure 3-4). When comparing the ICE curve estimation to the observed species ri chness, a little more than one third of firefly fauna has been estimated to be collected. The ICE curve estimated 673.45 species are yet to be found in the country of Colombia; the observe d species richness was 273. Of these 273 species, most are likely to be undescribed. Zaragoza-Caballero et al (2003) found that after one year of collecting in a tropical dry forest their observed species richness wa s 76% (19 observed : 25.4 estimated) of the ICE estimation of firefly species richness. Their study was conducted in the Sierra de Huautla Biosphere Reserve in Mexico (600 km2) a tropical dry forest. Tropical dry forests are some of the most diverse habitats in Central America (Janzen, 1988), and as such this comparatively low observed species richness is probably due to low numbers of traps being used (six per month for five days as opposed to an average of 87.6 traps per month for two weeks at a time) and the shorter time frame of the study (one year as opposed to four years). The Zaragoza-Caballero et al (2003) value for a Mexican dry forest can be compared to similarly sized area in Colombia, PNN Gorgona, to illustrate the vast diversity of Colombian fireflies even when compared to a similarly diverse tropical habitat. The PNN Gorgona is

PAGE 45

45 roughly 616 km2 and is located in the southwest corn er of Colombia on the Pacific coast. Compared to the Sierra de Huautla Biosphe re Reserve (SHBR), PNN Gorgona has around 75% more observed species (SHBR = 19, PNN Gorgona = 32). The PNN Gorgona estimated species richness was an order of magnitude highe r than SHBR (SHBR = 25.4, PNN Gorgona = 114.28). The estimated species diversity is close to four fold higher than the observed species diversity for PNN Gorgona, and more than four fold higher than the estimated species diversity of SHBR. The diversity of Colombian fireflie s in this single area, both esti mated and observed, is enormous when compared to a similarly sized area in a much different habitat.

PAGE 46

46 CHAPTER 4 FIREFLY ABUNDANCE AND SPECIES RICHNESS ACR OSS ELEVATION AND THROUGH TIME Introduction The abundance of insects, both in term s of i ndividuals and species, has been found in most groups to vary with respect to time and elevation (Lawton, et al ., 1987; Fleishman et al., 1998; Wolda et al ., 1998; Lees et al., 1999). Wolda et al (1998) define seasonality as a pattern of variation that repeats itself over successive years. This variation of seasonal patterns of species richness is common in temperate areas and is es pecially pronounced alon g latitudinal gradients (Wolda et al ., 1998). Seasonality has also been found to be common among insects in tropical areas where the climate is relativ ely stable, though seasonal pattern s of variation are still evident (Dobzhansky and Pavan, 1950; Bigger, 1976). Not onl y species richness but ecological functions such as trophic structure among coleopteran a ssemblages have been shown to differ between years (Noriega et al ., 2007). Seasonality diluted interspe cific competition for resources among Amazon beetles in Colombia, which may have been a driving factor for the diversification of this coleopteran feeding guild. However, northern Pa namanian mosquitoes have been found to not exhibit variation between years when climatic variables were considered (Wolda and Galindo 1981). The vast majority of Neot ropical insect taxa have not been examined for effects of seasonality on their distribution. The interplay between elevation and seasonality has also received little attention in Neot ropical insects. These aspects of biological diversity require much more research before larger patterns and trends across taxa can be described. As an example of tropical seas onality in insects, the season al diversity and abundance of Neotropical weevils in Panama was examined for periods of between .5 ye ars to three years at seven collection sites from 1974 to 1984 (Wolda et al ., 1998). A total of 2030 weevil species were collected over this time period, with only 30.1% of the species collected having been

PAGE 47

47 previously described. These seas onal patterns of speci es richness were observed for all sites studied even in those sites which were relativel y climatologically stable. For comparison, studies of weevil species richness in Pale arctic temperate areas found ric hness values that ranged from 29 species in a Danish beech forest (Nielson, 1974) up to 128 species in a Polish Oak-hornbeam forest (Witkowski, 1975). Zaragoza-Caballero et al (2003) found that firefly species richness in Sierra de Huautla Biosphere Reserve in Mexico, a tropical dry forest ranging in elev ation from 700 m to 2400 m above sea level, varied over time within th eir one year study. Because the time frame of the study did not include multiple years, the seasonal ity of this group was not addressed. However, they do comment that the seasonal variation in species richness seemed to be mainly due to nutritional resource availability. The samples for the study at hand were collected over a much wider range of habitats, includi ng tropical rain forests, co astal mangrove forests and many others, and over a period of four years. Much work, both empirical and theoretical, has been done proposing potential causes for frequency distributions of species ric hness over elevational gradients. Lawton et al (1987) summarize four possibilities for the decrease in species diversity with increasing altitude: reduced habitat area; reduced resource divers ity; more severe and va riable environmental conditions; and reduced primary productivity at higher elevations Rahbek (1995) proposed that the once over-generalization of monotonically decreasing specie s diversity with increasing elevation was mainly due to incorrectly citing as well as over citation of two studies: Terborgh (1977) and Kikkawa and Williams (1971). The m onotonic decrease in species richness with elevation can often be compensated for by standardizing the amount of area within each elevation. Terborgh (1977) showed that by standardizing the area of each elevation, a

PAGE 48

48 humpbacked curve was generated near the mid-point of eleva tion in his study. Rahbek claims that a horizontal or humpbacked curve is likely to be the most typical for patterns of species richness over elevation. The designation of what constitutes mid elev ation for a study is often an arbitrary assignment. However, mid elevati on would most correctly be defi ned as the segment of the study area which occupies the middle third of elevatio n variation. So, if a study area included a range of elevation from 0 3000 m above sea level, the mid elevation segment would be from 1001 m to 2000 me above sea level. A mid-domain effect (MDE) occurs when th e random placement of species geographic ranges on a bounded map produces a peak of speci es richness near the center. (Colwell et al ., 2004) Mid-domain effects are models that place the highest species richness at the center of the geographic region. The MDE is used as a null m odel. A null model attempts to explain the pattern of variation without using the factor or mechanism of intere st (in this case the effect of being placed in the middle of a geographic range ) while using as many other factors as possible (Colwell and Winkler, 1984). This null model is then compared to a model which includes the factor of interest so as to analyze the contributio n of the factor on the model produced. Middomain effects have been shown to explain, with variable success, the pe rcentage of variation between observed species ri chness and that predicted by an MDE model. Colwell et al (2004a) presented data on 21 empirical studies of MDEs, with percent variation explained from 0% to 96%, with the average being 54.91%. The utility of MDEs is very often affected by the taxa being examined. Different types of taxa have varying degrees of success when modeled using MDEs. Colwell et al (2004) found that modeling insect taxa using MDEs did not differ from all other taxa being examined (13% 91% variation explained, average of 48.2%).

PAGE 49

49 The correlation of seasonality (months of the years of study) and elev ation with individual and species abundance for firef lies will here be calculated fo r the collection sites within Colombia. A long-term study of firefly seasonality as well as species fr equency distributions over elevational gradients has not been carried ou t for any geographic region. It is hoped that the results of this study will help to stimulate future research focused on the seasonality of Neotropical insects as well as to assist in im proving the efficiency and accuracy of future sampling protocols. Materials and Methods Specimens The specim ens used in this study were collected from January 2000 to December 2003. In total, 276 morphospecies and 2029 individuals we re processed and reco rded from Colombia. Statistical Methods The range of elevation w as divided into bi ns of 100 m each, beginning at sea level and continuing to 3900 m (Sanders, 2002). The total nu mber of morphospecies and individuals for each was bin was calculated. Each morphospeci es and individual was assumed to occur throughout the entire range of elevati on for the bin (or bins) it was found in. Traps set during this study were typically placed in the field for two weeks. The time period for this study was divided into monthly bins, beginning in January 2000 and continuing to December 2003. For each month, the total numbe r of morphospecies and individuals was calculated. If a trap that captured fireflies was set between two months, then those individuals and morphospecies were assumed to occur throughout both months. The correlation of species with time, species across elevation, indivi duals with time, and individuals across elevation were calculated using the Data Analys is add-in for Microsoft Excel.

PAGE 50

50 The significance of these values was tested using a least-squares linear re gression, also using the Data Analysis add-in for Microsoft Excel. Results Statistical Methods: F or both individuals and species there was a negative correlation between advancing time and abundance of indivi duals and species collected from January 2000 to December 2003 (individuals: coe fficient of variation = -4.106, R2 = 0.4432; species: coefficient of variation = -1.405, R2 = 0.5053) (Figure 4-1). A least-squares linear regression analysis showed that there was a significant co rrelation in each case (i ndividuals: df = 1, var = 7456.41, r2 = 0.443, p < 0.0001; species: df = 1, var = 733.691, r2 = 0.419, p < 0.0001). For both individuals and species there wa s a negative correlation between increasing elevation and abundance of individuals and sp ecies caught from January 2000 to December 2003 (individuals: coefficient of variation = -0.7895, R2 = 0.0475; species: coeffici ent of variation = 0.1813, R2 = 0.0903) (Figure 4-2). A least-squares linear regression analysis showed that the correlation was significant only in the case of species and increasing elevation (df = 1, var = 134.059, r2 = 0.075, p = 0.026) while the correlation betwee n individuals and increasing elevation was not signifi cant (df = 1, var = 4834.21, r2 = 0.052, p = 0.0647). Discussion These analyses show the distribution of fire fly diversity and individual abundance vary with respect to tim e, supporting th e findings of Zaragoza-Caballero et al (2003) that cantharoid diversity varied with time. The time period of this study coincided with three oceanic temperature events: two La Nia s (January to June, 2000; October, 2000 to February, 2001) and one El Nio (May, 2002 to March, 2003). Accordi ng to the National Weather Service, a Nio/a

PAGE 51

51 Figure 4-1. Morphospecies and i ndividuals caught through the course of the study. The red line in dicates individuals and the b lue line indicates morphospecies.

PAGE 52

5 2 Figure 4-2. Morphospecies and i ndividuals caught across el evations of Colombia. The red line indicates individuals and the blu e column indicates morphospecies.

PAGE 53

53 event occurs when the Oceanic Nio Index (ONI) is .5C, or more severe, for three months or more for the average oceanic temperature in 5 N-5S, 120-170W geographic region (National Weather Service, 2008). A La Nia event occurs when the ONI is -0.5C or less, and an El Nio event occurs when the ONI is +0.5C or more. If the ONI is plotted along with the number of individuals and morphospecies co llected, a pattern emerges. More fireflies, both in terms of individual abundance and species richness, were collected du ring periods when the ONI was increasing, and fewer individual and species were caught during La Nia events (Figure 4-3). During the El Nio events, which are periods of warm ocean temperatures, South America especially along the Pacific coast receives warmer and much wetter summers (Trenberth, 1997); conversely, La Nia events, which are periods of cool ocean temperatures, cause South America especially along the Pacific coast to ha ve cooler and much drier summers. The MaxEnt analysis completed in chapter three demonstrated that the predicted distribution for fireflies within Colombia was mo st affected by the driest and coolest quarters, which together accounted for 68.4% of the variability within the model. The findings from the MaxEnt analysis concerning te mperature and precipitation values also match the pattern shown here in relation to La Nia and El Nio events ; the greatest abundance of species and individuals were captured during periods of where the ONI was increasing. Duri ng months where ONI values were low (-0.5 or lower) a reduced quantit y of both individuals an d morphospecies were collected (January to March, 2000; November, 2000 to January 2001) when compared to the immediately preceding or following months. For most of Colombia, the rainy season occu rs during May through October (Figure 4-4). It was during the rainy season of 2000 that the greatest abundance of fireflies (both individuals and morphospecies) was collected. This finding supports Zaragoza-Caballero et al .s (2003)

PAGE 54

5 4 Figure 4-3. Morphospecies and i ndividuals caught through the course of the study with the ONI index plotted. The red line indi cates individuals, the blue indicates species, and the orange line indicates the ONI value. The dashed lines demarcate the area of the graph between which average ONI values occur, .5C

PAGE 55

55 findings that the greatest divers ity of cantharoids (including fire flies) was collected during the rainy season. However, the periods increased individual and morphospecies abundance occurred during the periods of December, 2000 to February, 2001 and December, 2001. These periods occurred during or prior to la rge increase in the ONI index as seen in Figure 4-3. Differences in trapping effort through the c ourse of the study could account for variation seen in abundance of individuals and morphospeci es collected (Figure 4-5). Trapping effort, as defined by the number of traps set over a time period, fluctuated around an average of 121.75 traps per month through the period of March 2000 to December 2001. After this period a sharp decline in trapping effort continued until January 2003, when effort increased to an average of about 79.92 traps per month. This sharp decline in trapping effort could explain why so few individuals and morphospecies were caught during this time. As trapping effort began to increase in 2003, the abundance of individuals and morphospecies did not increase accordingly. This coul d be due to an oversight on the part of the researchers, as one group of samples (ranging fr om specimens collected from 2001-2004) did not get sorted for inclusion in the study. While these samples are currently bei ng sorted, the range of those samples across years does not appear to provide a negative bi as against any specific time range. The results of this study suggest that th e abundance of fireflies within Colombia is predicted to increase during the dry and cool pe riods, while periods of lowest individual and morphospecies abundance occur during or following a strong El Nio event (Figure 4-3). This period of very wet rainy seasons could have cause d the fireflies habitats to become flooded and unusable by them. This, coupled with the decrea sed trapping effort, could explain the low abundance of individuals and mo rphospecies collected during th e period following December, 2002.

PAGE 56

5 6 Figure 4-4. Morphospecies and individuals caught through the c ourse of the study with the ti me period of rainy seasons superimposed. The red line indicates individu als, the blue indicates species, the gr een regions indicate months of the rainy season for the majority of the geographic range of Colombia

PAGE 57

5 7 Figure 4-5. Morphospecies and i ndividuals caught through the course of the study with the number of traps plotted. The red lin e indicates individuals, the blue indicates species, and the green line indicates the total number of malaise traps set during that month.

PAGE 58

58 Tests for correlation across the elevational ra nge of Colombia showed that the abundance of morphospecies collected decr eased as elevation increased. The relationship was not strong, and the graph shows a loose bimodal distribution with respect to morphospecies collected, with the highest peak ranging be tween 0 1000 m and the second, lower peak ranging from 2801 3900 m. This is distinctly different than the di stribution of traps set, with the highest peak between 0 200 m and the second highest peak between 1800 2000 m. The curve representing traps set tends to approximate the individuals and morphospeci es collected curve until 700 m (Figure 4-6). At this elevati on in the PNN Farallones de Cali park, a very large series of Psilocladus sp. were collected (more than 200 individuals ). This park, which spans the elevations of 710 900 m, had the greatest amount of specime ns collected. In addition, the trapping effort through this elevational range yiel ded a high proportion of individua ls and species per trap when compared to the majority of other ranges (Figure 4-6). Trapping effort was not equally effective at all elevations as can be seen in Figure 4-6. Trapping effort was most effective at lower el evations (<1000m) while in some areas of high trapping effort (1701 2100m, 280 1 2900m) there was a very low number of individuals and morphospecies caught per trap. Mala ise traps, the types used in this study, are typically used to capture fireflies. However, the focus of the trapping effort for this study was on an entirely likelihood of capturing parasitic hymenoptera, and the fireflies ca ught were incidental to the original study. The more likely scenario is that there are more firefly individuals and species at lower elevations resulting in a greater proportion of caught specime ns using an equal or lesser number of traps.

PAGE 59

5 9 Figure 4-6. Morphospecies and indi viduals caught and malaise traps used across elev ations of Colombia with the number of traps plotted. The red column indicates indivi duals, the blue column indicates species and the green line indicates the total number of malaise traps set over that elevation

PAGE 60

60 CHAPTER 5 GEOGRAPHICAL DISTRIBUTION OF CO L OMBIAN FIREFLY SEXUAL SIGNALING MODALITIES Introduction Fireflies are well known to the general public because m any species exhibit conspicuous bioluminescent flash patterns. However, not all fi refly species are bioluminescent as adults and therefore do not use photic sign als to attract or locate ma tes (McDermott 1964; Branham and Wenzel 2003.). Decelle (1945), Lloyd (1972), Lloyd et al (1989) and de Cock and Matthysen (2005) have demonstrated that se veral different species of firef lies use pheromonal sexual signals (Gorham, 1880; McDermott, 1911; Mast, 1912; McDermott, 1914; Hess, 1922; Blair, 1924; Ohba, 1983; David and Birch, 1989). Lloyd (1972) found that males of Lucidota atra (Say), Photinus indicus (Le Conte), and Pyropyga nigricans (Say) were attracted to petri dishes that both contained females and which had previously contained females. McDermott (1964) also discussed the possibility that many firefly spec ies could use a combination of both photic and pheromonal signals within a courtship context. Lloyd et al (1989) observed males of a Thai species of firefly (Pteroptyx valida ) using both photic and pher omone signals in its sexual signaling system. When in close proximity to a receptive female a male would place their terminalia in the face of the female; during th is display there would be little to no photic emissions. Studies by de Cock and Matthysen (2005) demonstrated that the firefly Phosphaenus hemipterus Fourcoy while being weakly luminescent, primarily used pheromonal signals, rather than photic, for sexual communication. Based upon their phylogenetic analysis of the family Lampyridae worldwide, Branham and Wenzel (2003) provided empirical evidence that pheromone only sexual signals were the most ancestral of the three signal systems (pheromone pheromone + photic, pho tic) within the family. Stanger-Hall et al (2007) constructed a molecular phyloge ny using three genes, 18s, 16s, and

PAGE 61

61 COI, of mostly N. American lampyrid taxa. The most parsimonious optimization of the three signaling modes on this phylogeny co rroborates the finding that the sole use of pheromones is the ancestral mode of si gnaling in Lampyridae. Specific signal modalities (e.g. pheromones vs. visual signals) might be more effectively dispersed through specific types of habitat (e .g. dense vegetation vs. open areas (Lloyd, 1979). No studies to date have empirically examined th e diversity of signal moda lities in firefly species across large geographical areas. Colombia is a geographically and ecologically diverse country (Wyngaarden and Fandio-Lozano, 2005). By analyzing sexual signaling modalities with regards to geographical distinctness, a new and inform ative trait of firefly diversity within ( -diversity), between ( -diversity) and among a ll considered sites ( -diversity) can be extrapolated. The distribution of firefly signaling modalities are here modeled across Colombia using the MaxEnt free software program. This is the first time that a predicted distribution model of firefly sexual signaling modalities has been created for any region. The diversity of the Colombian firefly fauna makes this an especially informative initial st udy of distributions of sexual signaling modalities. This study can help provide a better unders tanding of the underlying environmental parameters which affect the sexual signaling modalities, of each modality individually and collectively, of fireflies. Materials and Methods Specimens The lam pyrid specimens used in the study were acquired through The Colombia Arthropod Project. This project involved a collaborative effort between the Humboldt Institute (Villa de Leyva, Colombia), the University of Kentucky and the Natural History Museum of Los Angeles County with funding through th e National Science Foundation, (DEB 0205982). Specimens of the family Lampyridae were collected from site s across the country of Colombia from 2000 to

PAGE 62

62 2004 (Table 2-1), sorted from other taxa and shipped in EtOH to Dr. Marc Branham at the University of Florida where they were pinned and sorted to morphospecies. Determining Sexual Signaling System Three general types of signali ng m odalities were recognized: 1. Pheromonal systems where one sex releases a long range chemical signal to attract the opposite sex; 2. Pheromonal and Photic systems where either or both sexes can employ both signaling modalities in either the attracti ng of or locating a mate and 3. Photic systems where fireflies, most often males, signal for mates by producing species specific bioluminescent emission (usually flas hes) wherein the opposite sex responds with a similar bioluminescent emission. The specific signaling system of the firefly mo rphospecies studied in this analysis was determined by Dr. Marc Branham based upon his knowledge of firefly morphology and behavioral ecology, as well as his experience with Ne otropical fireflies. The general criteria used were: 1) Pheromonal signal system = pectinate, bipectinate, or serrate antennae; small eyes and lack of photic organ, 2) Pheromonal + Photic signal system = pectinate, bi pectinate, or serrate antennae; presence of photic organ, and 3) Photic signal system = filiform antennae, large eyes, and/or presence of a photic. While many of the mo rphospecies included in this study were not represented by both male and female specimens (Table 7-1) the criteria above appears appropriate for diagnosing general signal sy stem types (Branham and Wenzel 2003.) Distribution Modeling MaxEnt (ver. 3.1.0) was used to m odel the distribution of firefly sexual signaling modalities across Colombia. The MaxEnt software program uses the principle of maximum entropy (Philips et al ., 2004) to model distributions of taxa by acquiring the distribution of maximum extent that agrees with all of the e nvironmental parameters at the occurrence points

PAGE 63

63 but which does not make unnecessary assumptions or place unneeded constraints on the data (Pearson et al ., 2007). (Chapter 3 for a more detailed discussion of the MaxEnt software) Morphospecies were divided into three diffe rent sexual signaling system bins based upon their hypothesized system: pheromone use only, phe romone + photic signals used, photic signal use only. Twenty bioclimatic and elev ation data layers were downloaded from the WorldCLIM dataset (http://www.worldclim.org/) (Hijmans et al ., 2005) (Table 2-2 for explanation of layers). Thirty percent of the data was set aside for m odel testing and the MaxEnt program was set to run for a maximum of 10,000 iterations. Jackknifi ng of the data was selected so that the contributions of individual laye rs to the model could be quantified. The options Random Seed and No Duplicate presence localities we re used for the analysis (Elith et al ., 2006); the default settings for all other parameters were unchanged. Results Distribution of Morphospecies and Individuals in Different Sexual Signaling Modalities Fireflies whose sexual signaling m odality was pheromone-based comprised the majority of both morphospecies and individual s caught with greater than 50% of morphospecies and nearly 75% of individuals. Photic fireflies were the ne xt most abundantly caught, followed by fireflies that use both photic and pheromone signaling. (Table 5-1) Distribution Models The Pheromone signaling system (Figure 5-1) had a m aximum test sensitivity plus specificity logistic threshold of 0.006 (p = 5.827E-4) and this area included 50.8% of the area of the map. These threshold and p-values indicate that a very highly significant portion of the sites used for testing were included within this fractio nal area of the map. The rate of omission of test samples was less than 0.001, and the test AUC was 0.802 out of a maximum of 0.904. Data

PAGE 64

64 Table 5-1. Number of morphos pecies and individuals by se xual signaling system type Sexual signaling modality Number of morphospecies Number of individuals Percent of total Morphospecies Individuals Pheromone 135 1494 51.72 73.63 Photic + Pheromone 51 263 19.54 12.96 Photic 75 272 28.74 13.41 layers of precipitation of coldest quarter (66.1 %), precipitation of drie st quarter (11.8%) and precipitation seasonality (6.9 %) were found to be the jac kknifed variables of greatest contribution. The Pheromone + Photic signaling system (Figure 5-2) had a maximum test sensitivity plus specificity logistic thres hold of 0.214 (p = 4.87E-5) and this area accounted for 19.1% of the area of the map. The combination of the threshold a nd p-values for this analysis show that a very highly significant portion of the s ites used for testing were included within this fractional area of the map. Test omission rate was less than 0.001, and the test AUC was 0.849 out of maximum possible 0.913. The jackknifed variables of greatest contribution were precipitation of coldest quarter (37.4%), altitude (17.9%), and precipitation of drie st month (15.3%). Finally, the Photic signaling system (Figure 5-3), the maximum test sensitivity plus specificity logistic threshold was 0.478 (p = 7.633E-7) and this area included 13.3% of the map area. The threshold value and corresponding p-valu e mean that a very highly significant portion of the testing sites were included within this area. Rate of omission of test samples was less than 0.001. The test AUC was 0.913 out of a possible 0.854. It has been found that with data that is drawn directly from the MaxEnt distribution the maximum achievable AUC is often less than one (Philips et al ., 2006). Also, the test AUC may exceed the maximum (Philips et al ., 2006). The jackknifed variables of great est contribution were precipitation of coldest quarter with 70.7% variation contributed, altit ude with 10.6% contributed varia tion, and precipita tion of driest month with 10.0% vari ation contributed.

PAGE 65

65 Figure 5-1. Log-likelihood map of predicted pheromone signaling morphospecies distribution for Colombia and immediately surrounding ar eas. Warm colors (i.e. red) indicate a higher likelihood of occurrence. White squa res indicate training points, purple squares indicate test points

PAGE 66

66 Figure 5-2. Log-likelihood ma p of predicted photic + pher omone signaling morphospecies distribution for Colombia and immediately surrounding areas. Warm colors (i.e. red) indicate a higher likelihood of occurrence. White squares indicate training points, purple squares indi cate test points

PAGE 67

67 Figure 5-3. Log-likelihood map of predicted photic signaling morphospecies distribution for Colombia and immediately surrounding areas. Warm colors (i.e. red) indicate a higher likelihood of occurrence. White squa res indicate training points, purple squares indicate test points

PAGE 68

68 Discussion The log-likelihood distribution of each sexua l signaling system (pheromone, pheromone + photic, and photic) was distinct although areas of overlap did occur. Fireflies whose hypothesized sexual signaling system was photic had large areas of very high probability (p 0.9 1) throughout the Andean re gion of Colombia as well as areas of high probability on the northwest Pacific coast (p 0.8). The areas of highest probability corresponded to glaciated landscapes of open grasslands and floodplains of riparian forests (Wyngaarden and FandioLozano, 2005). The Amazonian midlands had larg e areas of intermediate probability (p 0.4 0.6) which corresponded to sedimentary plains of evergreen forests (Wyngaarden and FandioLozano, 2005). Lall et al (1980) found that certain temperate species of flashing fireflies emitted wavelengths of light that were adapted to be maximally perceived during time periods where reflectance from green foliage was high. These fire flies were adapted to areas of open habitat where their flashes would be able to be perceived by females at a distance. Flashing fireflies would be at a reproductive disadvantage in hab itats such as dense jungle and forest canopies, where their flashes would be absorbed or obstruc ted by surrounding vegetati on as the direct line of sight between two individuals is greatly reduced. Colombian flashing fireflies have likely adapted to the same constraints as their temperat e counterparts in this way by being distributed in areas of open grassland and fore sts with sparse undergrowth. Those firefly species that used pheromone, onl y courtship signals had a distribution which closely mirrored the distribution of all fireflies throughout Colombia (Figure 3-1). This could be an artefact of sampling since the majority of fireflies caught (> 73.6%) had a hypothesized pheromone sexual signaling system. However, th ere were no areas of very high log-likelihood probability (p 0.9 1). Areas of high probability (p 0.8) were found in the Amazonian midlands consisting of sedimentary plains of evergreen forests as well as along the southwest

PAGE 69

69 Pacific coast consisting of coastal plains of mangrove forests (Wyngaarden and Fandio-Lozano, 2005). Pockets of moderate probability (p 0.7) were found in the Andes corresponding to glaciated landscapes of open grasslands (Wyngaarden and Fandio-Lozano, 2005). Pheromone releasing fireflies do not appear to be as affect ed by dense vegetation as are photic fireflies. Fireflies whose hypothesized sexu al signaling system is a co mbination of photic signals and pheromone signals had a distri bution that was localized much more than either photic or pheromone only systems. This reduced log-likel ihood probability distribution was located in similar areas to both the photic and pheromone signalers. However, the extent of very high probability (p 0.9 1) was increased, encompassing a wider area. These areas were along the central Andean region of Colombia as well as th e southwest Pacific coast. Areas of low to high probability (p 0.5 0.8) were greatly reduced, especially in the Amazonian midlands. Also, of all three signaling types, th e log-likelihood probability dist ribution of photic + pheromone signaling fireflies was the only one that was visi bly affected by the Amazon River (Figure 5-4). Photic + pheromone signaling fireflies had a dr astically reduced log-likelihood probability (p 0.3 p 0 0.1) over the Amazon River as well as having lower probabilitie s near the river (p 0.1 0.3 in pheromone + photic compared to p 0.3 0.6 in photic or pheromone). All three distributions had data layer BIO 19, precipitation of coldest quarter, as the greatest contributor to the model. Pheromone si gnalers differed from the others, having data layers BIO17, precipitation of dr iest quarter, and BIO 15, precipitation seasona lity, as the second and third most important layers; both photic and photic + pheromone signalers had ALT, altitude, and BIO14, precipitation of driest month. Pheromone fireflies seem to be more affected by longer dry seasons and more fluctuating patte rns of precipitation than photic or photic + pheromone fireflies.

PAGE 70

70 A B C Figure 5-4. Log-likelihood distributions for sexual signaling modalities for areas surrounding the Amazon River. A) Pheromone, B) Photic + pheromone, and C) Photic. Warm colors (i.e. red) indicate a higher likeli hood of occurrence. White squares indicate training points, purple square s indicate test points

PAGE 71

71 Altitude was one of the three most important variables of importance for both fireflies that have pheromone + photic sexual signaling moda lity and photic fireflies. Pheromone + photic signaling fireflies had the greatest diversity of species in areas of low elevation (<300 m) and mid elevation (600 900 m) (Figure 5-5). Photic signaling fireflies also had the greatest diversity below in these same ranges. The distribution of pheromone signaling fire flies species diversity was also most diverse at lower elevations (<300 m). However, firef lies with a pheromone signaling modality had large spikes in diversity in mid elevations (600 900 m) as well as in higher elevations (>1800 m). Pheromonally signali ng fireflies have been able to successfully utilize a wider variety of habitats and have grea ter diversity over a wide r range of elevations. The differences between the sexual signali ng modalities distribution patterns and the processes found to affect the variation in their distribution show that each of the three signaling modalities occupy different areas of geographic space. There are general areas of geographic overlap between the three systems, but the dis tinct log-likelihood distribution of probabilities between them show that each signaling modality is most likely to found in areas that are distinct to that modality.

PAGE 72

7 2 Figure 5-5. Distribution of number of species with different sexua l signaling modalities across elevation.

PAGE 73

73 CHAPTER 6 CONCLUSIONS The three objectives for this study were to: 1. analyze the geographic distribution and species rich ness of fireflies within Colombia and elucidate the variables most affecting their current distribution, 2. estimate th e species diversity of Colombia and areas within the country both temporally and spatially at geographic and elevational scales, and 3. examine the distribution of firefly sexual signaling modalities throughout Colombia. The results of each objective helped to provide support for several co nclusions: fireflies are not randomly distributed throughout the geographic and spatial range of Colombia; there are environmental parameters that correlate with current firefly distributions; these environmental parameters differentially affect the current distributions of the three se xual signaling modalities; and in comparison to a previous study Colombian firef lies are much more diverse. Objective 1 The firefly fauna of Colom bia was found to be very diverse. The Colombian firefly fauna includes more than 10% (273) of the estimated number of world firefly species, although this figure will likely decrease with the continued desc ription of new firefly species. This is most aptly demonstrated by the species accumulation cu rve obtained in Chapter 3 (Figure 3-4), where less than 50% of the firefly species have b een estimated to be collected. The Fishers and Evenness values for Colombia as a whole ( 84.9 and 0.7345) indicate th at the country has a relatively even distribu tion of individuals per morphospecies There were a few morphospecies with a large abundance of specimens caught (> 10 0), but the Evenness value indicates that the distribution of species was sti ll relatively even across the different areas and throughout morphospecies proportions.

PAGE 74

74 The Lampyridae as a whole have a predicted di stribution defined by relatively few factors, with the amount of precipitation (during the drie st and coldest quarters) being most important. Altitude was also found to be a contributing factor to the distribution of fireflies. Predicted Colombian firefly distribution correlated well with the two regions of Colombia designated as world biodiversity hotspots: the tropi cal Andes and the Choc (Myers, 2000). Objective 2 Firefly diversity was highest below 1000 m (fi gure 4-2), and the jackknifing of variable importance showed that altitude was one of the three variables that contributed most to the analysis. Although trapping effort was also highest over this range, the ratio of fireflies collected, both individuals and morphospecies, to traps set was highest in this ra nge. Increased trapping effort at higher elevations did not yield result s even remotely comparable to those at lower elevations, even when many traps were se t (figure 4-2, 600 900 m compared to 1800 2000 m). The MaxEnt analysis for the fireflies as a whole showed that alti tude was an important variable for the predicted firefl y distribution. This was corrobora ted by the correlation analysis and least squares linear regression which show ed that as elevation increased both species richness and abundance of individuals decreased although only the correla tion between elevation and species was shown to be signi ficant. However, the MaxEnt di stribution maps predicted that the areas with the greatest log-likelihood of fire fly distribution were areas in the Andes with elevations greater than 1 000 m. This could be due to the fact that the MaxEnt software chooses a percentage of samples at the beginning of the analysis with which to create the log-likelihood distribution map. Although the program does reach the optimal maximum entropy distribution with this percentage of the sample, the true di stribution of all of the samples input into the program could very well be different. Although Elith et al (2006) found that satisfactory

PAGE 75

75 distribution models could be cal culated with as few as five samples using MaxEnt, these satisfactory distribution models are neither reliable nor excellent. An improvement could be made to the MaxEnt software package that woul d include an option to sample, with or without replacement, a specified number of times the enti rety of the specimens chosen for the analysis and then average the predictions of these analyses to create a more realistic log-likelihood distribution. This would likely consume a much larger amount of computer processing power than the present version of MaxEnt. However, this is an area of theoretical and practical research using MaxEnt that should be explored. Objective 3 Each of the three sexual signaling modalities wa s affected by a com bination of factors. All three sexual signaling modalities were affected by a period of dry weather, ranging from the driest quarter (pheromone) to the single driest month (pheromone + photic, photic). Pheromone + photic fireflies and photic fireflies were also hi ghly affected by altitude, whereas pheromone fireflies were not as affected by altitude. This was corroborated in figure 5-5, which showed that pheromone fireflies, while having the highest diversity below 1000 m, also had pockets of diversity at higher elevations (2800 to 3400 m). Pheromone firef lies with few exceptions made up 50% of the firefly species at elevations above 1800 m. Regard less of sexual signaling type, the greatest diversity of firefly morphospecies was found below 1000 m. Fireflies that use pheromone + photic si gnals had a predicte d distribution which corresponded very closely to the areas of Colo mbia that are designate d as world biodiversity hotspots (Myers et al ., 2000). To a lesser extent photic fire flies exhibited a similar distribution, but their predicted log-likelihood values for thes e areas are lower (~0.5 0.7 for photic fireflies opposed to ~0.7 1.0 for pheromone + photic firefl ies). These areas include the tropical Andes and the Choc region along the Pacific coast. As designated world biodiversity hotspots (Myers

PAGE 76

76 et al ., 2000), these areas are among the most threaten ed in the world as well as being among the most diverse. Although fireflies utilizing these two sexual signaling modalities do not have the highest species richness (combined they cont ributed less than 50% of the morphospecies diversity of Colombian fire flies) they are the type of firefly that is most charismatic, at least to the general public. As such, these areas could emphasize their high likelihood of flashing firefly occurrence to highlight the importance of conser vation of arthropods and invertebrates in these hotspots. The species richness distribution of sexual signaling modalities of Colombian fireflies differs from that of North America (Marc Branham, personal communication). In North America, there is a greater richness of flashing firefly species, whereas in Colombia there are nearly twice as many pheromonally signaling fireflies as there are photically signaling fireflies. Comparisons and Concluding Remarks The Zaragoza-Caballero et al (2003) study was conducted in the Sierra de Huautla, biosphere reserve in Mexico consisting of a tropical dry forest of around 600 km2. As was shown before, PNN Gorgona, a park within Colombia of approximately equal area, had a higher observed and much higher estimated species richness values (19 to 32, 25.4 to 114.28). For comparison, a much smaller area, Reserve Nacional La Planada (32 km2) had an observed species richness slightly less than Sierra de Hu autla (17 compared to 19) but had an estimated richness almost 50% higher (36.37 to 25.4). This ar ea is classified as pluvial premontane forest (Holdridge, 1979), receiving an average of 4087 mm of rain per year. The species accumulation curves for this area were also different from the larger park Zaragoza-Caballero et al (2003) studied, with neither the observed or estim ated species accumulation curve reaching an asymptote, although neither was rising e xponentially as was PNN Gorgona curves.

PAGE 77

77 This discrepancy between Zaragoza-Caballero et al .s (2003) previous study and the findings reported here highlights the rich biodi versity of fireflies within Colombia. Fireflies have been hypothesized to be pa rticularly rich within the Neotropical region (Lloyd, 2002) and this study provides evidence that supports this hypothesis. The knowledge base of Neotropical fireflies, both taxonomically and systematically, is in desperate need of attention, as is evidenced by the fact that few of our specimens were ev en able to identified with reliable generic designations. The forests of Colomb ia are rapidly disappearing (Etter et al 2006a, 2006b) and with them a large portion of not only firefly dive rsity but the biodiversity of all life in the region that includes such distantly related organisms as b acteria and birds. It is hoped that the findings here can help direct future researchers and cons ervation scientists to more effectively allocate and acquire appropriate land for the cons ervation of these and other organisms.

PAGE 78

78 APPENDIX A MAXENT MORPHOSPECIES DISTRIBUTION MAPS MaxEnt log-likelihood dist ributions were created for all possible m orphospecies. Morphospecies collected in less than five unique localities were unable to have a distribution created. For all morphospecies 30 % of the samples were used fo r model testing, a maximum of 10000 iterations was allowed, and the options No Duplicate presence records and Random Seed were used. All other opti ons were left default. Generic identifications, when possible, were made using Greens (n.d.) key to the genera of world.

PAGE 79

79 A B C D Figure A-1. Log-likelihood map of predicted distributions for selected morphospecies of Lucidota in Colombia and immediat ely surrounding areas. A) Lucidota sp. 1, B) Lucidota sp. 2, C) Lucidota sp. 3, D) Lucidota sp. 4. Warm colors (i.e. red) indicate a higher likelihood of occurrence. White squa res indicate training points, purple squares indicate test points.

PAGE 80

80 A B C D Figure A-2. Log-likelihood map of predicted distributions for selected morphospecies of Lucidota in Colombia and immediat ely surrounding areas. A) Lucidota sp. 6, B) Lucidota sp. 7, C) Lucidota sp. 8, D) Lucidota sp. 9 in. Warm colors (i.e. red) indicate a higher likelihood of occurre nce. White squares indica te training points, purple squares indicate test points.

PAGE 81

81 A B C D Figure A-3. Log-likelihood map of predicted distributions for selected morphospecies of Lucidota Ledocas and Macrolampis in Colombia and immediately surrounding areas. A) Lucidota sp. 10, B) Lucidota sp. 11, C) Ledocas sp. 1, D) Macrolampis sp. 1. Warm colors (i.e. red) in dicate a higher likelihood of occurrence. White squares indicate training points, purple squares indicate test points.

PAGE 82

82 A B C D Figure A-4. Log-likelihood map of predicted distributions for selected morphospecies of Photuris in Colombia and immediat ely surrounding areas. A) Photuris sp. 1, B) Photuris sp. 2, C) Photuris sp. 3, D) Photuris sp. 9. Warm colors (i.e. red) indicate a higher likelihood of occurrence. White squa res indicate training points, purple squares indicate test points.

PAGE 83

83 A B C D Figure A-5. Log-likelihood map of predicted distributions for selected morphospecies of Photinus and Phaenolis in Colombia and immediat ely surrounding areas. A) Photinus sp. 2, B) Photinus sp. 5, C) Photinus sp. 6, D) Phaenolis sp. 1. Warm colors (i.e. red) indicate a higher likelihood of occurrence. White squares indicate training points, purple squares indicate test points.

PAGE 84

84 A B C D Figure A-6. Log-likelihood map of predicted distributions for selected morphospecies of Dryptelytra Pygolampis and Tenaspis in Colombia and immediately surrounding areas. A) Dryptelytra sp. 1, B) Dryptelytra sp. 3, C) Pygolampis sp. 1, D) Tenaspis sp. 1. Warm colors (i.e. red) indicate a hi gher likelihood of occurre nce. White squares indicate training points, purple squares indicate test points.

PAGE 85

85 A B C D Figure A-7. Log-likelihood map of predicted distributions for selected morphospecies of Magnoculus in Colombia and immediately surrounding areas. A) Magnoculus sp. 1, B) Magnoculus sp. 2, C) Magnoculus sp. 3, D) Magnoculus sp. 4. Warm colors (i.e. red) indicate a higher likeli hood of occurrence. White squa res indicate training points, purple squares indicate test points.

PAGE 86

86 A B C D Figure A-8. Log-likelihood map of predicted distributions for selected morphospecies of Magnoculus and Psilocladus in Colombia and immediately surrounding areas. A) Magnoculus sp. 5, B) Magnoculus sp. 6, C) Psilocladus sp. 1, D) Psilocladus sp. 2. Warm colors (i.e. red) in dicate a higher likelihood of occurrence. White squares indicate training points, purple squares indicate test points.

PAGE 87

87 A B C D Figure A-9. Log-likelihood map of predicted distributions for selected morphospecies of Psilocladus in Colombia and immediat ely surrounding areas. A) Psilocladus sp. 3, B) Psilocladus sp. 5, C) Psilocladus sp. 6, D) Psilocladus sp. 7. Warm colors (i.e. red) indicate a higher likelihood of occurrence. White squares indicate training points, purple squares indicate test points.

PAGE 88

88 A B C D Figure A-10. Log-likelihood map of predicted di stributions for selected morphospecies of Psilocladus and unidentified morphospecies in Colombia and immediately surrounding areas. A) Psilocladus sp. 10, B) 6, C) 24, D) 32. Warm colors (i.e. red) indicate a higher likelihood of occurrence. White squares indicate training points, purple squares indicate test points.

PAGE 89

89 A B C D Figure A-11. Log-likelihood map of predicted distributions for selected unidentified morphospecies in Colombia and immediat ely surrounding areas. A) 33, B) 34, C) 35, D) 39. Warm colors (i.e. red) indicate a higher likelihood of occurrence. White squares indicate training points, pur ple squares indicate test points.

PAGE 90

90 A B C D Figure A-12. Log-likelihood map of predicted distributions for selected unidentified morphospecies in Colombia and immediat ely surrounding areas. A) 40, B) 41, C) 43, D) 44. Warm colors (i.e. red) indicate a higher likelihood of occurrence. White squares indicate training points, pur ple squares indicate test points.

PAGE 91

91 A B C D Figure A-13. Log-likelihood map of predicted distributions for selected unidentified morphospecies in Colombia and immediat ely surrounding areas. A) 45, B) 46, C) 47, D) 58. Warm colors (i.e. red) indicate a higher likelihood of occurrence. White squares indicate training points, pur ple squares indicate test points.

PAGE 92

92 A B C D Figure A-14. Log-likelihood map of predicted distributions for selected unidentified morphospecies in Colombia and immediat ely surrounding areas. A) 59, B) 61, C) 73, D) 76. Warm colors (i.e. red) indicate a higher likelihood of occurrence. White squares indicate training points, pur ple squares indicate test points.

PAGE 93

93 A B C D Figure A-15. Log-likelihood map of predicted distributions for selected unidentified morphospecies in Colombia and immediat ely surrounding areas. A) 77, B) 78, C) 87, D) 88. Warm colors (i.e. red) indicate a higher likelihood of occurrence. White squares indicate training points, pur ple squares indicate test points.

PAGE 94

94 A B C D Figure A-16. Log-likelihood map of predicted distributions for selected unidentified morphospecies in Colombia and immediat ely surrounding areas. A) 89, B) 90, C) 93, D) 94. Warm colors (i.e. red) indicate a higher likelihood of occurrence. White squares indicate training points, pur ple squares indicate test points.

PAGE 95

95 A B C D Figure A-17. Log-likelihood map of predicted distributions for selected unidentified morphospecies in Colombia and immediat ely surrounding areas. A) 95, B) 98, C) 99, D) 126. Warm colors (i.e. red) indicate a higher likelihood of occurrence. White squares indicate training points, pur ple squares indicate test points.

PAGE 96

96 A B C D Figure A-18. Log-likelihood map of predicted distributions for selected unidentified morphospecies in Colombia and immedi ately surrounding areas. A) 128, B) 129, C) 130, D) 132. Warm colors (i.e. red) indicate a higher likelihood of occurrence. White squares indicate training points, pur ple squares indicate test points.

PAGE 97

97 APPENDIX B SPECIES ACCUMULATION CU RVES FOR SITES REPR ESENTED BY FIREFLY SPECIMENS Observed (Sobs) and estimated (ICE) species accumulation curves were calculated for each area that fireflies were collected in. Each datase t was analyzed using the EstimateS program. A maximum of one thousand iterations was allowed, each dataset was randomized without replacement, and the Fishers Shannon, and Simpson indices were calculated. All other options were left default.

PAGE 98

98 A B Figure B-1. Species accumulation curves for se lected sites (1, 2). A) SFF Iguaque, B) PNN Gorgona. Solid line is observed (Sobs Mao Tau) species rich ness; dashed line is estimated (ICE) species richness.

PAGE 99

99 A B Figure B-2. Species accumulation curves for se lected sites (3, 4). A) PNN Amacayacu, B) PNN Tayrona. Solid line is observed (Sobs Mao Tau) species rich ness; dashed line is estimated (ICE) species richness.

PAGE 100

100 A B Figure B-3. Species accumu lation curves for selected sites ( 5, 6). A) RN La Planada, B) PNN Sierra Nevada de Santa Mart a. Solid line is observed (Sobs Mao Tau) species richness; dashed line is estimated (ICE) species richness.

PAGE 101

101 A B Figure B-4. Species accumulati on curves for selected sites (7, 8). A) PNN Serrana de Chiribiquete, B) PNN El Tuparro. Solid line is observed (Sobs Mao Tau) species richness; dashed line is estim ated (ICE) species richness.

PAGE 102

102 A B Figure B-5. Species accumulation curves for selected sites ( 12, 13). A) PNN Utria, B) PNN Chingaza. Solid line is observed (Sobs Mao Tau) species rich ness; dashed line is estimated (ICE) species richness.

PAGE 103

103 A B Figure B-6. Species accumulation curves for se lected sites (15, 18). A) SFF Los Colorados, B) PNN Farallones de Cali. Solid line is observed (Sobs Mao Tau) species richness; dashed line is estimated (ICE) species richness.

PAGE 104

104 A B Figure B-7. Species accumulation curves for selected sites (22, 23). A) PNN Sierra de la Macarena, B) PNN Cueva de los Gucharos. Solid line is observed (Sobs Mao Tau) species richness; dashed line is estimated (ICE) species richness.

PAGE 105

105 A B Figure B-8. Species accumulation curves for selected sites (24, 25). A) PNN La Paya, B) PNN Sumapaz. Solid line is observed (Sobs Mao Tau) species rich ness; dashed line is estimated (ICE) species richness).

PAGE 106

106 A B Figure B-9. Species accumulati on curves for selected sites (26, 30). A) Estacon Biolgica Mosiro-Itajura, B) PNN Los Katio s. Solid line is observed (Sobs Mao Tau) species richness; dashed line is estim ated (ICE) species richness.

PAGE 107

107 APPENDIX C TABLE OF SPECIES PRESENCE IN EACH SITE For each site the presence or absen ce of each mor phospecies is recorded. Each site is numbered according to Table 3-1. An X denotes that a mor phospecies was caught at that site, while a denotes that it was not caught at that site.

PAGE 108

108 Table C-1. Presence and absen ce of morphospecies by site Site 1 2 3 4 5 6 7 8 12 13 15 16 18 22 23 24 25 26 27 28 30 M.sp.1 X X X X M.sp.2 X M.sp.3 X X X X M.sp.4 X X M.sp.5 X X X X X X X X X X M.sp.6 X X M.sp.7 X X X M.sp.8 X M.sp.9 X M.sp.10 X M.sp.11 X M.sp.12 X X M.sp.13 X M.sp.14 X M.sp.15 X M.sp.16 X X X X M.sp.17 X X X X M.sp.18 X X X X X M.sp.19 X M.sp.20 X M.sp.21 X M.sp.22 X M.sp.23 X M.sp.24 X X M.sp.25 X X X M.sp.26 X X M.sp.27 X M.sp.28 X M.sp.29 X X M.sp.30 X M.sp.31 X X X M.sp.32 X X X X M.sp.33 X X X M.sp.34 X M.sp.35 X X M.sp.36 X M.sp.37 X X X X X X M.sp.38 X X X M.sp.39 X X M.sp.40 X X X X M.sp.41 X X M.sp.42 X M.sp.43 X M.sp.44 X M.sp.45 X X X M.sp.46 X X X

PAGE 109

109 Table C-1. Continued Site 1 2 3 4 5 6 7 8 12 13 15 16 18 22 23 24 25 26 27 28 30 M.sp.47 X M.sp.48 X X M.sp.49 X X X M.sp.50 X M.sp.51 X M.sp.52 X M.sp.53 X M.sp.54 X M.sp.55 X X M.sp.56 X X M.sp.57 X M.sp.58 X X X M.sp.59 X M.sp.60 X M.sp.61 X X M.sp.62 X M.sp.63 X M.sp.64 X X M.sp.65 X M.sp.66 X X M.sp.67 X M.sp.68 X M.sp.69 X M.sp.70 X M.sp.71 X X M.sp.72 X X M.sp.73 X X M.sp.74 X M.sp.75 X X M.sp.76 X X M.sp.77 X X X X M.sp.78 X M.sp.79 X M.sp.80 X M.sp.81 X X M.sp.82 X M.sp.83 X X M.sp.84 X M.sp.85 X M.sp.86 X M.sp.87 X X X M.sp.88 X X M.sp.89 X X X X M.sp.90 X X M.sp.91 X M.sp.92 X X X

PAGE 110

110 Table C-1. Continued Site 1 2 3 4 5 6 7 8 12 13 15 16 18 22 23 24 25 26 27 28 30 M.sp.93 X M.sp.94 X X M.sp.95 X M.sp.96 X X X M.sp.97 X M.sp.98 X M.sp.99 X M.sp.100 X M.sp.101 X M.sp.102 X M.sp.103 X M.sp.104 X M.sp.105 X M.sp.106 X M.sp.107 X M.sp.108 X M.sp.109 X M.sp.110 X M.sp.111 X M.sp.112 X M.sp.113 X X M.sp.114 X X M.sp.115 X M.sp.116 X M.sp.117 X M.sp.118 M.sp.119 X M.sp.120 X M.sp.121 X M.sp.122 X M.sp.123 X M.sp.124 X X M.sp.125 X X M.sp.126 X M.sp.127 X M.sp.128 X M.sp.129 X M.sp.130 X M.sp.131 X M.sp.132 X M.sp.133 X M.sp.134 X M.sp.135 X M.sp.136 X M.sp.137 X M.sp.138 X

PAGE 111

111 Table C-1. Continued Site 1 2 3 4 5 6 7 8 12 13 15 16 18 22 23 24 25 26 27 28 30 M.sp.139 X M.sp.140 X M.sp.141 X M.sp.142 X M.sp.143 X M.sp.144 X M.sp.145 X M.sp.146 X M.sp.147 X M.sp.148 X M.sp.149 X M.sp.150 X M.sp.151 X M.sp.152 X M.sp.153 X M.sp.154 X M.sp.155 X M.sp.156 X M.sp.157 X M.sp.158 X M.sp.159 X M.sp.160 X M.sp.161 X M.sp.162 X M.sp.163 X M.sp.164 X M.sp.165 X M.sp.166 X M.sp.167 X M.sp.168 X M.sp.169 X M.sp.170 X M.sp.171 X M.sp.172 X M.sp.173 X M.sp.174 X M.sp.175 X M.sp.176 X M.sp.177 X M.sp.178 X M.sp.179 X M.sp.180 X M.sp.181 X M.sp.182 X M.sp.183 X M.sp.184 X

PAGE 112

112 Table C-1. Continued Site 1 2 3 4 5 6 7 8 12 13 15 16 18 22 23 24 25 26 27 28 30 M.sp.185 X M.sp.186 X M.sp.187 X M.sp.188 X M.sp.189 X M.sp.190 X M.sp.191 X M.sp.192 X M.sp.193 X M.sp.194 X M.sp.195 X M.sp.196 X M.sp.197 X M.sp.198 X M.sp.199 X M.sp.200 X M.sp.201 X M.sp.202 X M.sp.203 X M.sp.204 X M.sp.205 X M.sp.206 X M.sp.207 X M.sp.208 X M.sp.209 X M.sp.210 X M.sp.211 X M.sp.212 X M.sp.213 X M.sp.214 X M.sp.215 X M.sp.216 X M.sp.217 X M.sp.218 X M.sp.219 X M.sp.220 X M.sp.221 X M.sp.222 X M.sp.223 X M.sp.224 X M.sp.225 X M.sp.226 X M.sp.227 X M.sp.228 X M.sp.229 X M.sp.230 X

PAGE 113

113 Table C-1. Continued Site 1 2 3 4 5 6 7 8 12 13 15 16 18 22 23 24 25 26 27 28 30 M.sp.231 X M.sp.232 X M.sp.233 X M.sp.234 X M.sp.235 X M.sp.236 X M.sp.237 X M.sp.238 X M.sp.239 X M.sp.240 X M.sp.241 X M.sp.242 X M.sp.243 X M.sp.244 X M.sp.245 X M.sp.246 X M.sp.247 X M.sp.248 X M.sp.249 X M.sp.250 X M.sp.251 X M.sp.252 X M.sp.253 X M.sp.254 X M.sp.255 X M.sp.256 X M.sp.257 X M.sp.258 X M.sp.259 X M.sp.260 X M.sp.261 X M.sp.262 X M.sp.263 X M.sp.264 X M.sp.265 X M.sp.266 X M.sp.267 X M.sp.268 X M.sp.269 X M.sp.270 X M.sp.271 X M.sp.272 X M.sp.273 X

PAGE 114

114 LIST OF REFERENCES Archangelsky, M., and M.A. Branham 2001. De scription of last instar and pupa of Pyropyga nigricans (Coleoptera: Lampyridae, Photinini) and comparison with larvae of other Photinini genera. Can. Entomol. 133: 155-164. ArcGIS (ArcMap). 9.1. 2005. Redlands, CA: Envir onmental Systems Resear ch Institute, Inc. Arnett, R.H., Jr., and M.C. Thomas. 2001. Intr oduction, pp. 1-17. In R.H. Arnett and M.C. Thomas (eds.), American Beetles, vol. 1. CRC Press, Boca Raton, FL. Barber, H.S. 1951. North American fireflies of the genus Photuris Smithsonian Inst. Collect. 117: 1-58. Beattie, A.J., and I. Oliver. 1995. Reply from A.J. Beattie and I. Oliver. Trends Ecol. Evol. 10: 203-204. Bigger, M. 1976. Oscillations of tropical insect populations. Nature 259: 207-209. Blair, K.G. 1924. Some notes on the luminosity of the Lampyridae. Trans R. Entomol. Soc. London, vii. Branham, M.A. 2008. Lampyridae. In R.G. Be utel & R.A.B. Leschen (eds.), Handbuch der Zoologie, Band IV Arthropoda: Insecta, Teilba nd 39, Evolution and Systematics. Waltyer de Gruyter, Berlin. [in press] Branham, M.A., and J.W. Wenzel. 2001. The evol ution of bioluminescence in cantharoids (Coleoptera: Elateroidea) Fla. Entomol. 84: 565-586. Branham, M.A., and J.W. Wenzel. 2003. The origin of photic behavior and the evolution of sexual communication in fireflies (Coleopt era: Lampyridae). Cladistics 19: 1-22. Brower, A.V.Z. 1995. Taxonomic minimalis m. Trends Ecol. Evol. 10: 203. Buck, J.B. 1937. Studies on the firefly. I. The eff ects of light and other agents on flashing in Photinus pyralis with special reference to periodicity and diurnal rhythm. Physiol. Zool. 10, 45-58. Bunge, J., and M. Fitzpatrick. 1993. Estimating the number of species: a review. J. Am. Stat. Soc. 88: 364-373. Campbell, I.C. 1995. Taxonomic minimalis m. Trends Ecol. Evol. 10: 203. Chao, A., W.-H. Hwang, Y.-C. Chen, and C.Y. Kuo. 2000. Estimating the number of shared species in two communities Stat. Sinica 10: 227-246. Chazdon, R.L., R.K. Colwell, J.S. Denlow, and M.R. Guariguata. 1998. Statistical methods for estimating species richness of woody regenera tion in primary and secondary forests of NE Costa Rica, pp.285-309. In F. Dallmeier and J.A. Comiskey (eds.), Forest

PAGE 115

115 biodiversity research, monitoring and mode ling: conceptual background and Old World case studies. Parthenon Publishing, Paris. de Cock, R., E. Mattysen. 2005. Sexual communication by pheromones in a firefly, Phosphaenus hemipterus (Coleoptera: Lampyridae) Anim. Behav. 70: 807-818. Colwell, R.K. 2005. EstimateS: statistical estimati on of species richness and shared species from samples. Version 8.0.0. Users Guid e and application published at: http://purl.oclc.org/estimates. Colwell, R.K., and J.A. Coddington. 1994. Estimating terrestrial biodiversity through extrapolation. Philos. T. R. Soc. B. 345: 101-118. Colwell, R.K., C.X. Mao, and J. Chang. 2004a. Interpolating, extrapolating, and comparing incidence-based species accumulation curves. Ecol. 85: 2717-2727. Colwell, R.K., C. Rahbek, and N.J. Gotelli. 2004b. The mid-domain effect and species richness patterns: what have we learned so far? Am. Nat. 163: E1-E23. Colwell, R.K., and D.W. Winkler. 1984. A nu ll model for null models in biogeography, pp. 344359. In D.R. Strong, Jr., D. Simberloff, L.G. Ab ele, and A.B. Thistles (eds.), Ecological communities: conceptual issues and the evid ence. Princeton Univer sity Press, Princeton, N.J. Cowles, R.B. 1959. Zulu journal. Universi ty of California Pr ess, Berkeley, CA. Crowson, R.A. 1972. A review of the classificat ion of Cantharoidea (C oleoptera), with the definition of two new families, Cneoglossidae and Omethidae. Rev. Univ. Madrid. 21 3577. Currie, D., and J. Kerr. 2007. Testing, as oppos ed to supporting, the mid-domain hypothesis: a response to Lees and Colwell ( 2007). Ecol. Letters 10: E9-E10. David, C. T. and M.C. Birch. 1989. Pheromones and insect behavior. In: Insect Pheromones in Plant Protection (Ed. by A. R. Jutsum & R. F. S. Gordon), pp. 17. New York: J. Wiley. Decelle, J. 1945. Communication lass.mens. du 3.II. 1945. Bull. Ann. Soc. Entomol. Bel. LXXXI, 36. Delong, D.C. II. 1996. Defining biodiversity. Wildlife Soc. B. 24: 738-749. Derraik, J.G.B., G.P. Closs, K.J.M. Dickinson, P.Sirvid, B.I.P. Barratt, and B.H. Patrick. 2002. Arthropod morphospecies versus taxonomic species: a case study with Araneae, Coleoptera, and Lepidoptera Conserv. Biol. 16: 1015-1023. Dobzhansky, T., and C. Pavan. 1950. Local and seas onal variations in relative frequencies of species of Drosophila in Brazil. J. An im. Ecol. 19: 1-14.

PAGE 116

116 Elith, J., C.H. Graham, R.P. Anderson, M. Dudi k, S. Ferrier, A. Guisan, R.J. Hijmans, F. Huettmann, J.R. Leathwick, A. Lehmann, J. Li, L.G. Lohmann, B.A. Loiselle, G. Manion, C. Moritz, M. Nakamura, Y. Nakazawa, J.M. Overton, A.T. Peterson, S.J. Phillips, K. Richardson, R. Scachetti-Pereira, R.E. Schapire, J. Sobern, S. Williams, M.S. Wisz, and N.E. Zimmermann. 2006. Novel methods improve prediction of species distributions from occurren ce data. Ecography 29: 129-151. van Erkel, A.R., and P.M.Th. Pattynama. 1998. Receiver operating characteristic (ROC) analysis: Basic principles and applications in radiology. Eur. J. Radiol. 27: 88-94. Erwin, T.L. 1988. The tropical forest canopy. Th e heart of biotic diversity. pp. 123-129. In Wilson, E.O. (ed.), Biodiversity. Nati onal Academy Press, Washington, D.C. Etter, A., C. McAlpine, S. Phinn, D. Pulllar and H. Possingham. 2006a. Unplanned land clearing of Colombian rainforests: spreading like a disease? Landscape Urban Plan. 77: 240-254. Etter, A., C. McAlpine, S. Phinn, D. Pulllar, and H. Possingham. 2006b. Characterizing a tropical deforestation wave: a dynamic spatial analysis of a deforestation hotspot in the Colombian Amazon. Glob. Change Biol. 12: 1409-1420. Ficetola, G.F., W. Thuiller, and C. Miaud. 2007. Prediction and validation of the potential global distribution of a problematic alien invasive species the American bullfrog. Diversity Distrib. 13: 476-485. Fisher, R.A., A.S. Corbet, and C.B. Williams. 1943. The relation between the number of species and the number of individuals in a random sa mple of an animal population. J. Animal Ecol. 12: 42-58. Fleishman, E., G.T. Austin, and A.D. Weiss. 1998. An empirical test of Rapoports rule: elevational gradients in montane but terfly communities. Ecology 79: 2482-2493. Gaston, K.J. 1991. The magnitude of global insect species richness. Conserv. Biol. 5: 283-296. Gibson, L., B. Barrett, and A. Burbidge. 2007. D ealing with uncertain absences in habitat modeling: a case study of a rare ground-dwel ling parrot. Diversity Distrib. 13: 1-9. Goldstein, P.Z. 1997. How many thi ngs are there? A reply to Oliv er and Beattie, Beattie and Oliver, Oliver and Beattie, and Oliver and Beattie. Conserv. Biol. 11: 571-574. Goldstein, P.Z. 1999a. Clarifying the role of species in ecosystem management: a reply. Conserv. Biol. 13: 1515-1517. Goldstein, P.Z. 1999b. Functional ecosystems a nd biodiversity buzzwords. Conserv. Biol. 13: 247-255. Gorham, H.S. 1880. On the structure of th e Lampyridae, with reference to their phosphorescence. Trans. R. Entomol. Soc. London, 63-67.

PAGE 117

117 Gotelli, N., and R.K. Colwell. 2001. Quantifying biodiversity: procedures and pitfalls in the measurement and comparison of specie s richness. Ecol. Lett. 4: 379-391. Green, J.W. Date unknown. [Key to worldwide Lampyridae genera, unfinished]. Unpublished generic key. Guisan, A., C.H. Graham, J. Elith, F. Huettmann, M. Dudik, S. Ferrier, R. Hijmans, A. Lehmann, J. Li, L.G. Lohmann, B. Loisell, G. Manion, C. Moritz, M. Nakamura, Y. Nakazawa, J.M. Overton, A.T. Peterson, S.J. Phillips, K. Richardson, R. Scachetti-Pereira, R.E. Schapire, S.E. Williams, M.S. Wisz, a nd N.E. Zimmermann. 2007. Sensitivity of predictive species distribution models to change in grain size. Diversity Distrib. 13: 332340. Hernandez, P.A., C.H. Graham, L.L. Master, and D.L. Albert. 2006. The effect of sample size and species characteristics on performance of different species distribution modeling methods. Ecography 29: 773-785. Hess, W.N. 1922. Origin and develo pment of the light organs of Photuris pennsylvanica De Geer. J. Morphol. 36, 245-277. Hijmans, R.J., S.E. Cameron, J.L. Parra, P.G. Jones, and A. Jarvis. 2005. Very high resolution interpolated climate surfaces for global land areas. Int. J. Climatol. 1965-1978. Holdridge, L.R. 1979. Ecologa basada en zonas de vida. Instituto Interamericano de Ciencias Agricolas, San Jos, Costa Rica. Hortal, J., P.A.V. Borges, and C. Gaspar. 2006. Ev aluating the performance of species richness indicators: sensitivity to sample gr ain size. J. Anim. Ecol. 75: 274-287. Jaramillo, M.A. 2006. Using Piper species diversity to identify conservation priorities in the Choc region of Colombia. Biodivers. Conserv. 15: 1695-1712. Janzen, D. 1988. Tropical dry forest. The most endangered major tropical ecosystem, pp. 130137. In Wilson, E.O. (ed.), Biodiversity. Na tional Academy Press, Washington, D.C. Kattan, G.H. 1992. Rarity and vulnerability: the bi rds of the Cordillera Central of Colombia. Conserv. Biol. 6: 64-70. Kerr, J.T., M. Perring, and D.J. Currie. 2006. Th e missing Madagascan mid-domain effect. Ecol. Lett. 9: 149-159. Kattan, G.H., P. Franco, C.A. Saavedra-Rodrguez, C. Valderrama, V. Rojas, D. Osorio, and J. Martnez. 2006. Spatial components of bird diversity in the Andes of Colombia: implications for designing a regional reserve system. Conserv. Biol. 20: 1203-1211. Kikkawa, J. and E.E. Williams. 1971. Altitude dist ribution of land birds in New Guinea. Search 2: 64-65.

PAGE 118

118 Kobori, H., and R.B. Primack. 2003. Part icipatory conservation approaches for satoyama the traditional forest and agricultural landscape of Japan. Ambio 32: 307-311. Krell, F.T. 2004. Parataxonomy vs. taxonomy in biodiversity studies pitf alls and applicability of morphospecies sorting. Biodiv. Conserv. 13: 795-812. Labella, D., and J.E. Lloyd. 1991. Lampyridae (C antharoidea), pp. 427-428. In F.W. Stehr (ed.), Immature insects, vol. 2. Kimball Hunt, Dubuque, IA. Lall, A.B., H.H. Seliger, and W.H. Biggley, J.E. Lloyd. 1980. Ecology of colors of firefly bioluminescence. Science 210: 560-562. Lawrence, J.F. and A.F. Newton. 1995. Families a nd subfamilies of Coleoptera (with selected genera, notes, references and da ta on family-group names. In J. Pakaluk and S.J. Slipinski (eds.), Biology, phylogeny, and classification of Coleoptera: papers celebrating the 80th birthday of Roy A. Crowson. Muzeum I In stytut Zoologii PAN, Warszawa, p. 849-861 Lawton, J.H., M. MacGarvin, and P.A. Heads. 1987. Effects of altitude on the abundance and species richness of insect herbivores on bracken. J. Anim. Ecol. 56: 147-160. Leather, S.R., and A.D. Watt. 2005. Sampling Theory and Practice, pp. 1-15. In S.R. Leather (ed.), Insect Sampling in Forest Ecos ystems. Blackwell Publishing. Oxford, UK. Lee, S., and A. Chao. 1994. Estimating population size via sample coverage for closed capturerecapture models. Biometrics 50: 88-97. Lees, D.C., and R.K. Colwell. 2007. A strong Ma dagascan rainforest MDE and no equatorward increase in species richness: reanalysis of The missing Madagascan mid-domain effect, by Kerr, J.T., Perring, M. & Currie D.J. (Ecol. Lett. 9: 149-159, 2006). Ecol. Lett. 10: E4E8. Lees, D.C., C. Kremen, and L. Andriamampianina. 1999. A null model for species richness gradients: bounded range overlap of butterflies and other rainforest endemics in Madagascar. Biol. J. Linn. Soc. 67: 529-584. Lloyd, J.E. 1964. Notes on flash communication in the firefly Pyractomena dispersa (Coleoptera: Lampyridae). Ann. Entomol. Soc. Am. 57: 260-261 Lloyd, J.E. 1965. Aggressive mimicry in Photuris: Firefly femme fatales. Science 149: 653-654. Lloyd, J.E. 1968. Illumination, another function of firefly flashes? Entomol. News 79: 265-268. Lloyd, J.E. 1971. Bioluminescent communication in insects. Annu. Rev. Entomol. 16: 97-122. Lloyd, J.E. 1972. Chemical communication in fi reflies. Environ. Entomol. 1: 265-266.

PAGE 119

119 Lloyd, J.E. 1979. Sexual selection in luminescent beet les. In M.S. and N.A. Blum (eds.), Sexual selection and reproductive competition in insects, pp. 293-342. Academic Press, New York, U.S.A. Lloyd, J.E. 2002. Lampyridae, pp. 187-197. In R.H. Ar nett, Jr., M.C. Thomas, P.E. Skelley, J.H. Frank (eds.), American Beetles, vol. 2. CRC Press. Boca Raton, FL. Lloyd, J.E., S.R. Wing, and T. Hongtrakul. 1989. Ec ology, flashes, and behavior of congregating Thai fireflies. Biotropica 21: 373-376. Loreau, M. 2000. Are communities saturated? On the relationship between and diversity. Ecol. Lett. 3: 73-76. Magurran, A.E. 2004. Measuring biological diversity. Blackwell Publ ishing. Oxford, UK. Mast, S.O. 1912. Behavior of fire-flies ( Photinus pyralis ) with special reference to the problem of orientation. J. An im. Behav. 2: 256-272. McDermott, F.A. 1911. A comparative study of the structure of the photogenic organs of certain American Lampyridae. Am. Nat. 45: 306-313. McDermott, F.A. 1914. The ecologic relations of the photogenic function among insects. Z. Wiss. Insekt. 10, 303-307. McDermott, F.A. 1958. The fireflies of Delawa re. Soc. Nat. Hist., Delaware, Wilmington. McDermott, F.A. 1964. The taxonomy of the Lam pyridae (Coleoptera). Trans. Am. Entomol. Soc. 90: 1-72. McDermott, F.A. 1966. Lampyridae. In W.O. Steel (ed.) Coleopterorum catalogus, pars 9: Lampyridae, 2nd ed., pp. 1-149. N.V. Drukker ij Hooiberg, Netherlands. McNeely, J.A. 2003. Conserving fore st biodiversity in times of violent conflict. Oryx 37: 142152. Metz, C.E. Basic principles of ROC analysis. Sem. Nuc. Med. 8: 283-298. Myers, N., R.A. Mittermeier, C.G. Mitterme ier, G.A.B. da Fonesca, and J. Kent. 2000. Biodiversity hotspots for conserva tion priorities. Nature 403: 853-858. National Weather Service, 2008. Cold and warm episodes by season. Retrieved July 23, 2008, from http://www.cpc.ncep.noaa.gov/products/analysi s_monitoring/ensostuff/ensoyears.shtml. Nielson, B.O. 1974. Studies of weevils (Curculionid ae) of a Danish beech forest (Curculionidae). Entomol. Medd. 42: 169-188.

PAGE 120

120 Noriega, J.A.A., J.P. Botero, M. Viola, and G. Fagua. 2007. Dinmica estacion al de la estructura trfica de un ensamblaje de Coleoptera en la Amazonia Colombiana. Rev. Col. Entomol. 33: 157-164. Ohba, N. 1983. Studies on the comm unication system of Japanese fi reflies. Sci. Rep. Yokosuka City Mus. 34: 17-23. OKeefe, S.T. 2002. Phengodidae, pp. 181-186. In R. H. Arnett, Jr., M.C. Thomas, P.E. Skelley, J.H. Frank (eds.), American Beetle s, vol. 2. CRC Press. Boca Raton, FL. Oliver, I., and A.J. Beattie. 1993. A possible me thod for the rapid assessment of biodiversity. Conserv. Biol. 7: 562-568. Oliver, I., and A.J. Beattie. 1996a Invertebrate morphospecies as surrogates for species: a case study. Conserv. Biol. 10: 99-109. Oliver, I., and A.J. Beattie. 1996b. Designing a cost-effective inve rtebrate survey: a test of methods for rapid assessment of biodiversity. Ecol. App. 6: 594-607. Palmer, M.W. 1990. The estimation of species richness by extrapolat ion. Ecol. 71: 1195-1198. Palmer, M.W. 1991. Estimating species richness: th e second-order jackknif e reconsidered. Ecol. 72: 1512-1513. Papi, F. 1969. Light emission, sex attraction and male flash dialogues in a firefly, Luciola lusitanica (Charp.). Monit. Zool. Ital. (N.S.) 3: 135-184. Pape M., and P. Gaubert. 2007. Modeling ecological niches from low numbers of occurrences: assessment of the conservation status of poorly known viverrids (Mammalia, Carnivora) across two continents. Dive rsity Distrib. 14: 1-13. Pearson, R.G., C.J. Raxworthy, M. Nakamura and A.T. Peterson. 2007. Predicting species distributions from small numbers of occurren ce records: a test case using cryptic geckos in Madagascar. J. Biogeogr. 34: 102-117. Philips, S.J., M. Dudik, and R.E. Schapire. 2004. A maximum entropy approach to species distribution modeling. In Proceedings of the 21st International Conference on Machine Learning. Banff, Canada. Philips, S.J., R.P. Anderson, and R.E. Schapi re. 2006. Maximum entropy modeling of species geographic distributions. Ecol. Model. 190: 231-259. Pik, A.J., I. Oliver, and A.J. Beattie. 1999. Ta xonomic sufficiency in ecological studies of terrestrial invertebrates. Au stral. J. Ecol. 24: 555-562.

PAGE 121

121 Prendergast, J.R., R.M. Quinn, J.H. Lawt on, B.C. Eversham, and D.W. Gibbons. 1993. Rare species, the coincidence of diversity hotspot s and conservation strategies. Nature 365: 335-337. Rahbek, C. 1995. The elevational gradient of spec ies richness: a uniform pattern? Ecography 18: 200-205. Rees, C.J.C. 1983. Microclimate and the flying Hemi ptera fauna of a primary lowland rainforest in Sulawesi, pp. 121-136. In S.L. Sutton, T.C. Whitmore, A.C. Chadwick (eds.), Tropical rainforest: Ecology and management. Black well Scientific Publications. Oxford, England. Sanders, N.J. 2002. Elevational gradients in ant species richness: area, geometry, and Rapoports rule. Ecograpy 25: 25-32. Simpson, E.H. 1949. Measurement of diversity. Nature 163: 688. Sivinski, J.M. 1981. The nature and possible func tions of luminescence in Coleoptera larvae. Coleopts. Bull. 35: 167-179. Sobern, J., and J. Lorente. 1993. The use of speci es accumulation functions for the prediction of species richness. Cons erv. Biol. 7: 480-488. Stanger-Hall, K.F., J.E. Lloyd, and D.M. Hillis 2007. Phylogeny of North American fireflies (Coleoptera: Lampyridae): implications for the evolution of light signals. Mol. Phylogenet. Evol. 45: 33-49. Stockwell, D.R.B, and A.T. Peterson. 2002. Controlling bias in biodiversity data, pp. 537-546. In J.M. Scott, P.J. Heglund, M.L. Morrison, J.B. Haufler, M.G. Raphael, W.A. Wall, F.B. Samson (eds.), Predicting species occurrences: Is sues of accuracy and scale. Island Press, Washington, DC. Takeda, M., T. Amano, K. Katoh. and H. Higuc hi. 2006. The habitat requ irement of the Genjifirefly Luciola cruciata (Coleoptera: Lampyridae), a repr esentative endemic species of Japanese rural landscapes. Biodivers. Conserv. 15: 191-203. Terborgh, J. 1977. Bird species diversity on an Andean elevational gradient. Ecology 58: 10071019. Thompson, G.G., S.A. Thompson, P.C. Withers, and J. Frasher. 2007. Determining adequate trapping effort and species richness accumulation curves for environmental impact assessments. Austral. Ecol. 32: 570-580. Toti, D.S., F.A. Coyle, and J.A. Miller. 2000. A st ructured inventory of Appalachian grass bald and heath bald spider assemblages and a test of species richness estimator performance. J. Arachnol. 28: 329-345.

PAGE 122

122 Trenberth, K.E. 1997. The definition of El Nio. Bull. Am. Meterol. Soc. 78: 2771-2777 Ugland, K.I., J.S. Gray, and K.E.Ellingsen. 2003. The species accumulation curve and estimation of species richness. J. Anim. Ecol. 72: 888-897. Walther, B.A. and J.L. Moore. 2005. The concepts of bias, precision and accuracy, and their use in testing the performa nce of species richness estimators, with a literature review of estimator performance. Ecography 28: 815-829. White, R.E. 1975. Trend curves of the rate of species description for North American Coleoptera. Coleopts. Bull. 29: 281-295. Williams, V.L., E.T.F. Witkowski, and K. Balk will. 2007. The use of incidence-based species richness estimators, species accumulation curves and similarity measures to appraise ethnobotanical inventories from SouthA frica. Biodivers. Conserv. 16: 2495-2513. Witkowski, Z. 1975. The ecology and succession of weevils (Coleoptera, Curculionidae) on mowed meadows in the Zabierzw area. Stud. Nat. 12: 1-81. Wolda, H., C.W. OBrien, and H.P Stockwell. 1998. Weevil diversity and seasonality in tropical Panama as deduced from light-trap catches (Coleoptera: Curculi onoidea). Washington, D.C.: Smithsonian Inst. Press. Wolda, H., and P. Galindo. 1981. Population fluctua tions of mosquitoes in seasonal tropics. Ecol. Entomol. 6: 99-106. Wyngaarden, van W., and M. Fandio-Lozano. 2005. Mapping the actual and original distribution of the ecosystems and the chorological types for cons ervation planning in Colombia. Divers. Distrib. 11: 461-473. Zaragoza-Caballero, S., F.A. Noguera, J.A. Chemsak, E. Gonzlez-Soriano, A. RodrguezPalafox, E. Ramrez-Garcia, and R. Ayala. 2003. Diversity of Lycidae, Phengodidae, Lampyridae, and Cantharidae (Coleoptera) in a tropical dry forest region in Mxico: Sierra de Huautla, Morelos. Pan-Pac. Entomol. 79: 23-37.

PAGE 123

123 BIOGRAPHICAL SKETCH Bradley Sm ith was born in 1983, in Clarksville Tennessee, to Debra and John Smith. He became interested in science and natural histor y at a young age, being encouraged to read books about paleontology, space and general science by his parents. In high school he developed an interest in veterinary medicine, which he would pursue for his first two years as an undergraduate at Western Kentucky University. During the summer of 2004, Brad ley took part in a study abroad course in South Africa and helped with university resear ch and conservation projects in Kenya. It was then he decided his passion was research instead of medicine. In the fall semester of his junior year at Western Kentucky University he attended Dr. Keith Phil ips entomology course, sparking his interest in insects, beetles and belostomatids particularly. These interests were encouraged by Dr. Philips, who employed him as a student re searcher from February 2005 to August 2006 where he sorted, curated and identified beetles for Dr. Philips fr om Africa and a project in Wisconsin. In August 2006 Bradley began his graduate studies at the University of Florida. At the University of Florida Bradley expanded his interests to include species distribution modeling and biodiversity estimation. He met hi s wife, Sheri Anderson, a fellow entomologist, while taking classes and teaching during his first year at the University of Florida. In Septembter 2008, Bradley was hired by Dr. Ronald Cave at the Indian River Re search and Education Center in Fort Pierce to be his seni or biological scientis t working to control the Mexican bromeliad weevil, Metamasius callizona, with Lixadmontia franki a Tachinid fly. Bradley graduated in the spring of 2009 from the University of Florida with his Master of Science in entomology.