USF Libraries
USF Digital Collections

Effects of prescribed fire on the diversity of soft-dwelling arthropods in the University of South Florida Ecological Re...

MISSING IMAGE

Material Information

Title:
Effects of prescribed fire on the diversity of soft-dwelling arthropods in the University of South Florida Ecological Research Area, Tampa, Florida
Physical Description:
Book
Language:
English
Creator:
Bellanceau, Celina
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Insects
Burn
Sandhill
Longleaf pine
Dissertations, Academic -- Biology -- Masters -- USF   ( lcsh )
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: Leaf litter samples were used to study arthropod diversity in the University of South Florida Ecological Research Area. Arthropods were collected from different plots of different prescribed burn frequencies (1year, 2 year, 5 year, 7 year and no burn frequencies). Differential responses of arthropods to burn frequency were studied. It was expected that burn frequency would affect arthropod richness and diversity and that the pattern of diversity and richness would follow the Intermediated Disturbance Hypothesis. It was also expected that richness and diversity would be correlated with the amount of leaf litter present. Preliminary results suggest that there are short-term and long-term differences among arthropod communities in response to fire history and vegetation and that the amount of leaf litter may influence the distribution of arthropod taxa.
Thesis:
Thesis (M.S.)--University of South Florida, 2007.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: World Wide Web browser and PDF reader.
System Details:
Mode of access: World Wide Web.
Statement of Responsibility:
by Celina Bellanceau.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 42 pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001921336
oclc - 191047829
usfldc doi - E14-SFE0001886
usfldc handle - e14.1886
System ID:
SFS0026204:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001921336
003 fts
005 20080122133432.0
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 080122s2007 flu sbm 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001886
035
(OCoLC)191047829
040
FHM
c FHM
049
FHMM
090
QH307.2 (ONLINE)
1 100
Bellanceau, Celina.
0 245
Effects of prescribed fire on the diversity of soft-dwelling arthropods in the University of South Florida Ecological Research Area, Tampa, Florida
h [electronic resource] /
by Celina Bellanceau.
260
[Tampa, Fla] :
b University of South Florida,
2007.
3 520
ABSTRACT: Leaf litter samples were used to study arthropod diversity in the University of South Florida Ecological Research Area. Arthropods were collected from different plots of different prescribed burn frequencies (1year, 2 year, 5 year, 7 year and no burn frequencies). Differential responses of arthropods to burn frequency were studied. It was expected that burn frequency would affect arthropod richness and diversity and that the pattern of diversity and richness would follow the Intermediated Disturbance Hypothesis. It was also expected that richness and diversity would be correlated with the amount of leaf litter present. Preliminary results suggest that there are short-term and long-term differences among arthropod communities in response to fire history and vegetation and that the amount of leaf litter may influence the distribution of arthropod taxa.
502
Thesis (M.S.)--University of South Florida, 2007.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
538
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
500
Title from PDF of title page.
Document formatted into pages; contains 42 pages.
590
Adviser: Ronald Sarno, Ph.D.
653
Insects.
Burn.
Sandhill.
Longleaf pine.
690
Dissertations, Academic
z USF
x Biology
Masters.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.1886



PAGE 1

Effects of Prescribed Fire on the Diversity of Soil-Dwelling Arthropods in the University of South Florida Ecological Research Area, Tampa, Florida by Celina Bellanceau A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science Department of Biology College of Arts and Science University of South Florida Co-Major Professor: Gary Huxel, Ph.D. Co-Major Professor: Ronald Sarno, Ph.D. Melissa Grigione, Ph.D. Gordon Fox, Ph.D. Date of Approval: February 15, 2007 Keywords: insects, burn, sandhill, longleaf pine Copyright 2007, Celina Bellanceau

PAGE 2

ACKNOWLEDGMENTS I would like to thank my Co-advisers Gary Huxel and Ronald Sarno for all their help and guidance during this process. For their valuable comments and advi ce I would like to thank Melissa Grigione and Gordon Fox, my committee members. I also thank Gary, Neal Halstead, Tere Albarracin, Tatiana Cornelisse n, Rebecca Forkner, Teresa Piacenza, Lance Arvidson and Drew Talley for their friendship an d all their help reviewing this thesis. For their help in the field I want to thank Tere, Teresa, Neal, and La nce and for their help with my analyses, I would like to thank Neal, Alison Meyers, Gary, Tatiana and Tere.

PAGE 3

i TABLE OF CONTENTS List of Tables................................................................................................................. .................................ii List of Figures................................................................................................................ ...............................iii Abstract....................................................................................................................... ....................................v Introduc tion................................................................................................................... ..................................1 Sandhill Co mmunity............................................................................................................. .......................1 Importance of fire............................................................................................................. ...........................1 Fire Effects on Sandhill Fl ora................................................................................................. .....................3 Importance of Soil Fauna....................................................................................................... .....................4 Fire Effects on Sandhill Litter and So il Fauna................................................................................. ...........5 Methods........................................................................................................................ ...................................7 Experimental Design............................................................................................................ .......................8 Data An alys is.................................................................................................................. ..........................12 Univariate Statis tics.......................................................................................................... .....................12 Multivariate Statis tics........................................................................................................ ....................13 Results........................................................................................................................ ...................................15 Overall........................................................................................................................ ...............................15 Dispersion..................................................................................................................... .........................16 Cluster an alys is............................................................................................................... ......................16 MDS............................................................................................................................ ..........................16 ANOSIM......................................................................................................................... ......................16 RELATE......................................................................................................................... .......................17 Repeated Meas ures ANOVA........................................................................................................ ........17 Seasonal....................................................................................................................... ..............................17 Cluster an alys is............................................................................................................... ......................17 MDS............................................................................................................................ ..........................17 ANOSIM......................................................................................................................... ......................17 Results of RELATE Spearma n rank correlation...................................................................................1 7 Traps.......................................................................................................................... ................................18 Plots.......................................................................................................................... .................................23 Traps.......................................................................................................................... ................................26 Seasonal....................................................................................................................... ..............................28 MDS............................................................................................................................ ..............................29 Two-Way Crossed AN OSIM: Seasonal............................................................................................... .....29 One Way ANOSIM: Seas onal....................................................................................................... ............30 Discussion..................................................................................................................... ................................31 Literature Cited............................................................................................................... ..............................34 Appendices..................................................................................................................... ...............................40 Appendix A: Dates of prescribed burns at the USF Eco Area...................................................................... .41 Appendix B: Dispersion Weighted and Square Root Transformed Si milarity matrix...................................42

PAGE 4

ii LIST OF TABLES Table 1. 15735 individuals used in the analyses. The data from the September 2004 sampling occasion were not included in the analyses because pitfa ll traps were not employed during that sampling occasion....................................................................................................................... ...................18 Table 2. Total and Mean Arthropod Abund ance per Fire Frequency............................................................18 Table 4. Leaf litter mass is found to increase from Ap ril 2005 to August 2005 from means of 97.34g to 126.76g. The leaf litter mass between August 2005 and November 2005 were similar.................18 Table 5. Arthropod Abundan ces per Fire Frequency............................................................................... .....19 Table 6. Mean Arthropod Abundan ces per Time Since Last Fire.................................................................19 Table 7. Mean Arthropod Order Richness per Fire Frequency.....................................................................1 9 Table 8. Mean Order Richne ss per Sampling Occasion............................................................................. ...19 Table 9. Total and Mean Leaf Litter Ma ss for each Plot and Sampling Occasion........................................20 Table 10. Mean Leaf Litter Mass per Fire Frequency............................................................................. ......20 Table 11. Mean Leaf Litter Mass per Fire Fre quency for each Samp ling Occasion.....................................21 Table 12. Mean Leaf Litter Ma ss per Time Sin ce Last Fire....................................................................... ...21 Table 13. Spearman Rank correlations for leaf litter mass and fire frequency for sampling occasions where leaf litter mass wa s recorded.................................................................................................. .........21 Table 14. Dispersion weighting is used to downweight highly variable orders. The null hypothesis of random dispersion is s upported if D = 1........................................................................................ .22 Table 15. Repeated measures analysis of variance (MANOVA) for differences within fire frequencies between sampling periods. Tests of Within-Subject s Effects indicate significant effects of season. ............................................................................................................................... .........................27 Table 16 Pairwise tests of sampling occasions across all fire frequ ency groups..........................................29 Table 17. Pairwise tests of sampling occasions........................................................................................... ..30 Table 18. Pairwise tests of fire fre quency and arthropod abundances...........................................................3 0

PAGE 5

iii LIST OF FIGURES Figure 1. Aerial photograph of the University of South Florida Ecological Research Area in northeast Tampa, FL. The experimental sandhill plots used in this study are outlined in yellow....................7 Figure 2. Map of the burn plots. By Emily Minor {, 2000 #142}. Based on GPS points of the outline of the plots; the areas are calculated by ArcView based on those data.......................................................8 Figure 3. The general set up of a pitfall trap. The tr aps used in this study contained antifreeze as the killing solution and did not have a hard ware cloth mesh covering...............................................................9 Figure 4. The presentation of a pitfall trap. The pitfall traps used in this study were 16oz plastic cups, filled 1/8 full of antifreeze. A two gallon bucket lid, propped up by sticks was used to shield the trap contents from su n and ra in..................................................................................................... ...........9 Figure 5. Tullgren funnel apparatus. The apparatus used in this experiment consisted of a beaker and a plastic funnel with ha rdware cloth screen...................................................................................... .10 Figure 6. Tullgren funnel set up for this experiment. Each 16 oz cup contained about an inch of 70% alcohol. Hardware cloth with mesh size 0.25 inch was used as a screen in the funnel. A 100 watt light bulb is placed directly over the funnel to drive ins ects down into the alcohol.......................11 Figure 7. Tullgren fu nnel trap station......................................................................................... ...................11 Figure 8. Cluster analysis dendrogram of arthropod co mmunity composition among experimental plots. Labels indicate plot. Solid lines indicate significan t differences among plots at p = 0.05. There are two major groups. One group consists of plots with the more extreme fire frequencies: 1 year and 7 years between fires and unburned plots. The other group consists of plots with more intermediate fire frequencies of 2 and 5 years between fires. All twoand five-year plots were burned one year prior to data collec tion, as well as the one-year plots. \ ........................................23 Figure 9. Cluster analysis dendrogram of experimental trap s. X-axis is labeled with traps. Solid lines indicate significant differences among plots at p = 0.05. Symbols indicate the amount of years since the last fire. There are two major groups. One group consists of the CWE trap, one of the traps that has not been burned for at least 30 years. The other group consists of all other traps with all the other times since last fire: 1 y ear, 7 year or at least 30 years...............................................24 Figure 10. Plot of nonmetric multidimensional scaling of experimental plots. Plot designations are labeled with colors and shapes. Dark blue triangles refer to plots with1 year fire frequencies, teal squares refer to plots with 2 year fire frequencies, red diamonds refer to plots with 5 year fire frequencies, purple circles refer to plots with 7 year fire fre quencies, and gray plus signs refer to the reference plots without a fire for at least 30 years. Contour lines indicate significantly different groups (SimProf at alpha = 0.05). The goodness-of-fit is high, indicated by the stress measure of 0.05. Higher goodness of fit is in dicated by lower stress.........................................................................25

PAGE 6

iv Figure 11. Plot of nonmentric multidimensional scaling of experimental traps. Trap designations are labeled with colors and shapes. Dark blue triangles refer to plots that were burned 1 year prior to data collection, teal squares refer to plots that had not been burned for 7 years prior to data collection, and red diamonds refer to plots that have not had a fire for at least 30 years prior to data collection. The goodness-of-fit is relatively low, indicated by the stress measure of 0.24. Lower goodness of fit is indicated by higher stress.............................................................................................. ..........26 Figure 12. Cluster analysis dendrogram of plots for eac h sampling occasion. X-axis is labeled with plots for each sampling occasions. Solid lines indicate significant differences among plots at p = 0.05. Plots seem to clump with other plots of the same sa mpling occasion with the exception of a few plots which are usually from an adjacent sampling occasion. Any outliers that are present are closest to the group of the same sampling occasion........................................................................................ 28 Figure 13. Plot of nonmetric multidimensional scaling of sampling occasions. Sampling occasion designations are labeled with colors and shapes The goodness-of-fit is relatively good, indicated by the stress measure of 0.13. Higher goodne ss of fit is indicated by lower stress.........................29

PAGE 7

v Effects of Prescribed Fire on the Diversity of Soil-Dwelling Arthropods in the University of South Florida Ecological Research Area, Tampa, Florida Celina Bellanceau ABSTRACT Leaf litter samples were used to study arthropod divers ity in the University of South Florida Ecological Research Area. Arthropods were co llected from different plots of different prescribed burn frequencies (1year, 2 year, 5 year, 7 year and no burn frequencies). Differential responses of arthropods to burn frequency were studied. It was expected that burn frequency would affect arthropod richness and diversity and that the pattern of diversity and richness would follow the Intermediated Disturbance Hypothesis. It was also expected that richness an d diversity would be correlated with the amount of leaf litter present. Preliminary results suggest that there are short-term and long-term differences among arthropod communities in response to fire history and vegetation and that the amount of leaf litter may influence the distribution of arthropod taxa.

PAGE 8

1 INTRODUCTION Sandhill Community Sandhill habitat is found in the temperate climate of th e southeastern United States in areas with deep, welldrained sandy soil deplete of nutrients and with almost no organic matter. The soils are marine deposits of once Plio-Pleistocene beach sand formations (Myers 1990). Sandhill is often called Longleaf pine/Turkey oak habita t because of the floral composition of the habitat. The overstory of sandhill habitat is mainly composed of Longleaf pine ( Pinus palustris ) with some slash pine ( Pinus elliotti ) while the understory is made up of turkey oak ( Quercus laevis ) and a few blackjack oaks ( Quercus incana ) or sand live oaks ( Quercus geminata ) (Schmidt 2005). Persimmon ( Diospyros Virginia ) and saw palmetto ( Serona repens ) also occur (Hay-Smith & Tanner 1997b). Underwood plant species can include sparkleberry ( Vaccinium arboretum ) and pawpaw ( Asimina incarna ) (Myers 1990). Although sandhill habitats have low diversity of trees, they do have high total diversity attributed to the species-rich herbaceous layer. Ground cover plants includes wiregrasses Aristida stricta and Sporobolus gracilis and can include other herbaceous plants such as lopsided Indian grass ( Sorghastrum secundum ), greeneyes ( Berlandiera pumila ), golden aster ( Chrysopsis scabrella ), blazing star ( Liatris pauciflora ), paint brush ( Carphephorus corymbosus ), and sandhill milkweed ( Asclepias humistrata ) as well as others (HaySmith & Tanner 1997b). Breeding amphibians use temporary ponds in sandhill habitat as a source of water that is free from predaceous fish and insects (Dodd & Charest 1988) and many fossorial animals require the loose sands of sandhill to “swim” or to burrow through (Myers 1990).The burrows of the Florida mouse ( Podomys floridanus ), gopher tortoises ( Gopherus polyphemus ), and the pocket gopher ( Geomys pineti ) serve as refuge for many other species, which also serve esse ntial roles in sandhills (Hay-Smith & Tanner 1997b). Other sensitive animals that utilize sandhill habitat include the Florida gopher frog ( Rana capito) the threatened eastern indigo snake ( Drymarchon corais ) and the Sherman’s fox squirrel ( Sciurus niger shermanii ) (Hay-Smith & Tanner 1997b). Gopher frogs an d the eastern indigo snakes are important commensals of gopher tortoise burrows. Scarab beetles such as Aephodius trogloytes Copris gopheri and Onthophagus polyphemi are some invertebrate commensals of gopher tortoise burrows (Myers 1990). Many of these animals help with seed dispersal, an d refugia and nest cavity construction (Hay-Smith & Tanner 1997b). Endangered species of sandhill habitat include red-cockaded woodpeckers ( Picoides borealis ) and Florida mice ( Podomys floridanus ) (Hay-Smith & Tanner 1997a), which w ill readily use gopher tortoise burrows instead of digging their own. Endangered plant speci es, which are endemic to sandhill, include clasping warea ( Warea amplexifolia ) and Carter’s mustard ( W. carteri ) (Liberton Undated). Importance of fire Sandhill communities are pyrogenic; the organisms living within it have evolved in the presence of fire, so that the habitat is fire-maintained and fire-depende nt (Agee 1993; Myers 1990). Sandhill requires frequent, low-intensity fires in one to ten year intervals to maintain its open overstory of mainly longleaf pines, understory of deciduous oaks and ground cover of wire grasses and forbs (Myers 1990).

PAGE 9

2 Fire often produces a mosaic of habitat (Mushinsky & Gibson 1991; Tansley 1924), which is important to biodiversity, including the diversity of arthropods. Co mplexity of habitat has been found to be positively associated with species richness (Hansen 2000; Humphrey et al. 1999; Huston 1994; Uetz 1979). The structure of the vegetation in an area influences popula tions of terrestrial arthropods. Different arthropods have specific microhabitat requirements and habitat heterogeneity allows for many different microhabitat types (Samways 2005). Fire has been found to be a significant abiotic agent of selection for many species and can be considered an ecological process (Agee 1993{Myers, 1990 #51)}. Fire can influence the structure of many communities as well as the evolution of the sp ecies within the commun ity. Fire can alter the size, age, and spatial distribution of habitats and the extent of the altera tion is correlated with fire frequency (Mushinsky & Gibson 1991). Of course, each plant community is different, so th e ecological effect of fire on habitat structure consequently differs. The plant species and litter of a community determine the temperature of a fire and the rate at which it spreads. Even within a community, the temperature and rate of spread is irregular and inconsistent which leads to the mosaic habitat patterns found in sandhill (Mushinsky & Gibson 1991). The lack of homogenous fire temperatures in a single stand of plants may be related to the type and height of vegetation (Hobbs & Gimingham 1984) and the subsequent moisture content of the litter layer (Hobbs et al. 1984) The plants living in sandhill provid e the fuel for the low intensity, fre quent fires that are necessary to maintain viable sandhill habitat. Sandhill and ponder osa pine communities produce well-aerated litter beds with their resinous needles which allow for incr eased flammability of the communities (Agee 1993). Lightning was historically the igniting force for sandhill fires and usually struck the lightning-rod like longleaf pines (Myers 1990). Native Americans have also ignited the fore sts for thousands of years (Kalisz 1986; Komarek 1974; Ware et al. 1989), so anthropogenic influences are nothing new to sandhill. It has been found that communities th at are exposed to any interval of fire are more similar to each other than to communities that are not exposed to fire (Hanula 2003). It has also been found that species richness and diversity can increase following a fire (Force 1981) Different genera differ in their response to fire, however, and it can seem to be an almost species-specific response (Ahlgren 1974; Hanula 2003; Niwa & Peck 2002; Siemann et al. 1997; Warren et al. 1987). Studies have been inconsistent in determining th e relationship between fire and the Intermediate Disturbance Hypothesis (IDH), which states that the highest diversity should be found in areas of intermediate disturbance. There are studies which have found that species richness increases with fire frequency, without an ensuing decline in species ri chness at higher fire frequencies (Kerstyn & Stiling 1999; Mehlman 1992; Nuzzo et al. 1996; Tester 1996). Other studies have found the lowest species richness to be at sites with intermediate disturbance (Schwilk et al. 1997), and others have actually found a decline in species richness with an increase in fire frequency (Collins et al. 1995). Another study found a decline in certain species when fire was suppress or occurred late (Andersen & Muller 2000). One reason for such inconsistent results could be the inclusion of dormant season fires, in some studies, which may impact fauna and flora differently than growing season fires (Beckage & Stout 2000). Differing time since last fire, which is known to affect species diversity could also be attributed to the mixed results found (Gibson & Hulbert 1987; Mehlman 1992). Beckage and Stout (2000) found no direct relationship between the number of fires and understory flowering stem species rich ness, diversity or abundance. They did fi nd an increased species richness and unimodal increase in flowering stem abundance and di versity resulting from fire, yet their model selection criterion favored the null model. Instead of a direct effect of fire, Beckage and Stout (2000) found indirect effects from open overstory. They found that overstory density and composition were most affected by varied fire frequency and this affected the understory composition.

PAGE 10

3 Reduced overstory density and composition impacts the composition and abundance of understory species by decreasing competition for light and soil resources Increased overstory density leads to decreased understory because of the light and soil resources that are being blocked or used. Support for this indirect relationship between disturbance and understory diversity was also found when using herbicides to reduce overstory density. Increased species richness and ground cover were also found to result from herbicidal declines in overstory oaks from a sandhill community (Brockway et al. 1998). Many studies have also found variation in responses to disturbance by arthropods. Schowalter (1994; 1995) found that Lepidopteran, detritivore and predator abundances decreased while sap-sucking insect abundances increased in response to canopy-opening disturbances such as hurricanes and timber harvesting. Studies on effects of pr escribed fire have also found th is trend for grasshopper (Kerstyn & Stiling 1999; Nagel 1973), leaf miner (Kerstyn & Stiling 1999) and herpetofaunal (Mushinsky & Gibson 1991); abundances were found to be higher in the more frequently burned sites. Reasons for the lack of support for the IDH could have been that the hypothesis does not operate in the Florida pine savanna system in which Beckage and Stout (2000) were studying. Another possibility could be that the relationship between the response variables and fire could be too subtle to detect with their experimental design, or there could have been low power attributed to the relatively small sample size of the study (Beckage & Stout 2000) Fire Effects on Sandhill Flora Disturbances such as fire are a selective force. The tim ing, frequency, and intensity of fire can have a huge effect on the species composition, abundances and the structure of sandhill communities. Fire is an abiotic disturbance that influences the habitat structure and, in turn, the evolution of stra tegies to survive. Species have evolved to be tolerant to fire in ecosystems with a history of fire (Mushinsky & Gibson 1991). Fire affects the germination of herbaceous plants, the regeneration of longleaf pines, the phenologies of forbs (Daubenmire 1968; Hartnett 1987; Platt et al. 1988), and the morphologies of flowering plants (Mushinsky & Gibson 1991; Platt et al. 1988). Longleaf pine is considered fire-resistant because it has evolved several adaptations to fire. Longleaf pines, like other fire adapted or resistant plants actually requi re fire to allow for the germination and establishment of its seedlings (Myers 1990(Agee 1993)). Seeds of longleaf pines need the open well-drained, coarsegrained soil to land on because the se eds are so large and have a persistent wing which cannot penetrate dense groundcover to reach the soil beneath. Seedlings require moderate shade, provided by palmettos and oaks, to protect them from the heat (Myers 1990). After a seedling has established itself it develops into its grass stage where it stays for two years to ten years. This grass stage provides it protection from fire because the terminal bud is nestled in the center of th e dense, long needles, whic h collect moisture. During the grass stage, the longleaf pine grows a long, thick taproot which provides energy stores for later bolting into a sapling. Bolting entails sending up a single straight stem that can grow up to one meter or more in a single growing season. The resulting sapling has a terminal bud which is likely to be above the fire line. Thick bark helps to insulate the tree and scaly plates are produced which flake off as they burn to dissipate heat. Thus, it is seen that longleaf pine communities requ ire frequent, low intensity fires to persist. If fire is disrupted for any reason and litter is allowed to accum ulate, recruitment could be decreased and a high temperature, high intensity fire could kill the longleaf pines (Myers 1990; Agee 1993). Sandhill habitats generally have groundcover consistin g of wiregrass, which allows for other groundcover such as forbs, low shrubs and other grasses to coexist. Some of these other groundcover species include bracken fern ( Pteridium aquilinum ), bluestems ( Andropogon spp.), dwarf live oak ( Q. minima ), golden aster and low-bush blueberry ( Vaccinium myrsinites ). Wiregrass is extremely sensitive to disturbances such as bedding, scraping and scarification of soils, and may not return to a site once it has been extirpated. Perplexingly, wiregrass has a low reproductive capacity yet it can often be found as a dense ground cover. Disturbance such as low intensity fire, grazing and minor soil disturbance during the growing season, April to mid-August, allows it bloom profusely (Myers 1990).

PAGE 11

4 After a fire, the degree of regeneration depends on the types of species present before the fire. Some plants have lignotubers which contain buds and energy stores, which allow for viable dormant buds below the surface. These buds are then stimulat ed to sprout by the loss of abov e ground shoots (Agee 1993). Habitats that have many fire tolerant plant species can resprout so rapidly that the it can return quickly to its pre-fire condition (Abrahamson 1984a, b; Keeley 1986; Malanson & Trabaud 1987). Plants with adventitious or latent axillary buds are able to regenerate by sprouting buds along the plant body (Agee 1993). This rapid return to the habitats original state leads to high productivity levels (Buchholz 1983). For many plant species, dispersal of seeds (Bradstock & Myerscoug h 1981(Agee 1993)) and flowering (Gill 1977; Knapp & Hulbert 1986; Wein & Bliss 1973) may increase followi ng a fire. A larger seed bank can result and may lead to new individuals becoming established in an area (Cheplick & Quinn 1987). Phenology and growth patterns may also change in species of plants enduring after a fire. Following a fire, new shoots, flowers and fruits may appear earlier during than they would if there had not been a fire. Burnt plants may also maintain their green color longer than plants that have not been burned (Daubenmire 1968). When fire is suppressed, less dominant sandhill sp ecies and invader species from adjacent upland communities take over, causing a decline in the fire-a dapted species a change in the overall vegetation. Because oaks thrive without fire, upland oak dominated sandhills can result if fire is suppressed. Upland oaks in the USF Eco Area are mainly turkey oaks and sand live oaks. These oaks are deciduous and generally of low stature and have multiple stems. Other hardwoods, such as persimmon also generally increase in abundance when fires are interrupted (Myers 1990). Importance of Soil Fauna Litter is composed of any decaying vegetative matter on the ground, such as leaves, woody material, fruit, and flowers. Soil and litter-dwelling arthropods ar e important in ecosystem processes such as decomposition of organic matter and the associated nutr ient cycling, horizontal and vertical movement of soil, and accumulation of soil which leads to soil stab ility in terrestrial ecosyste ms. Nutrient cycling is important and soil and litter-dwelling arthropods break down vegetative litter and decaying animal material including chitin into pieces that are more manageable fo r bacteria, protists and f ungi, which further degrade the matter (Eisenbeis & Wichard 1985; Gullan & Cr anston 2000; Hale 1976; Samways 2005; Wallwork 1967). Many arthropods also help in the distribution of fungi (Hale 1976; Wallwork 1967). Arthropods are found in just about every soil type in all parts of the world, (Gullan & Cranston 2000; Longino & Colwell 1997; Samways 2005; Wallwork 1976). They are small, fecund and respond quickly to environmental change and so are often used as subjects for studying biodiversity (Longino & Colwell 1997). Important arthropods of litter and humus-rich soils ar e acariformes (mites), isopte ran (termites), formicidae (ants), collembola (springtails), protura, diplura (d oubletails), thysanoptera (thrips), many coleoptera (beetles), as well as immature stages of coleoptera, diptera (flies), and lepidoptera (butterflies and moths). Collembola and Mites are major orders associated with soils and litter and are among the most abundant groups of soil fauna (Eisenbeis & Wichard 1985; Wallw ork 1976). Collembola can actually constitute up to 80% of total number of animals in the soil (Hale 1976). Soil arthropods that are restri cted to terrestrial soils include many blattodea (roaches), orthoptera (crickets and grasshoppers), and dermaptera (earwigs) (Gullan & Cranston 2000). Rather than being defined by geog raphic proximity, many communities of arthropods have been found to be best defined by habitat type (Hughes et al. 2000). Spatial scale is important to consider when discu ssing soil and leaf litter-dw elling arthropods because different morphs and species differ in their behaviors and habitat requirements (Chust et al. 2003). Some arthropods may be confined to a particular soil layer or depth and show morphological and behavioral adaptations specific to that layer or depth. Morphological adaptations can include larvae with welldeveloped legs for moving through soil and pupae with adaptations that allow them to move to the surface for emergence. Subterranean insects ca n also exhibit seasonal shifts in distribution. Drier layers are used for pupation and deeper soils, which offer constant temp eratures, are used more in the winter (Gullan & Cranston 2000; Hale 1976).

PAGE 12

5 Collembola and Mites often exhibit a vertical stratifi cation of species in leaf litter. Generally, organisms with larger body sizes are found in gr eater concentrations in the upper la yers of leaf litter, whereas smaller forms are found deeper. These organisms differ morphologically as well, with epedaphic organisms occurring closer to the surface having better developed eyes, springing organs (collembola), antennae and often more stronger pigmentation. Euedaphic organi sms are found deeper in the litter often are less pigmented and have reduced sensory and springing organs (Eisenbeis & Wichard 1985; Wallwork 1976). Most soil mites are mainly hemieda phic; living in the organic layer associated with the soil surface (Wallwork 1967). Seasonal changes in distribution have been seen for some organisms. Collembola have been found to peak in abundance in autumn with their lowest abundance in summer, although different species can vary in the timing of their peak abundances (Hale 1976). Arthropods vary in their feeding habits. Coleoptera can be predaceous, saprophagous, and coprophagous and their diets can vary depending on the life stage. Diptera can be parasitic, insectivorous, coprophagous, herbivorous, and saprophagous or scavengers. Mites can be predaceous, herbivorous or saprophagous (Raw 1967). Soil-dwelling arthropods including collembola, beetle larvae, certain termites and ants can be saprophages or detritivores and get their nutrition from eating large volumes of soil containing dead and decaying vegetable and animal debris, although it ma y be the soil-associated fungal and bacterial microorganisms which are being selected (Gullan & Cranston 2000; Raw 1967; Wallwork 1976). Microarthropod densities have been found to be higher close to rhizospheres; the rhizosphere of a plant is the zone including root hairs, mycorrhizae and rhizob acteria. Collembola, along w ith other microarthropods can be beneficial to plants because they aid in the transport of beneficial fungi to and bacteria and removal of harmful fungi and bacteria from the rhizosphere during selective grazing by the collembola on root and subterranean fungi, including pathogenic fungi. The interactions between microarthropods and fungi could make certain elements available for plants to u tilize because of the mineralization of nitrogen and phosphates by microarthropods, although this relationship is not fully understood (Gullan & Cranston 2000). There are many other potential causes of arthropod distributions including moisture content, temperature, light, and presence and distribution of resources or predators. McCoy (1987) found that Polopinus species were negatively correlated with reptilian predators of arthropods. Arachnids that are associated with soil and litter are spiders, mites and pseudoscorpions ( Pseudoscorpiones ). Spiders living in the soil zone are mostly tubular web spinning spiders such as Atypidae, Clubionidae, Eresidae, Agelenidae, Hahniid ae as well as migratory spiders such as jumping (Salticidae) and wolf spiders (Lycosidae) (Eisenbeis & Wichard 1985; Wallwork 1976). Although spiders and pseudoscorpions are strictly carnivorous (the latter mainly eating mites, collembola and nematodes) and are major predators of other soil fauna, there is a wide array of dietary groups for mites, including parasitism and predation (Wallwor k 1976). Predatory mites generally feed on the same prey as pseudoscorpions. Other predators of soil fauna include certain ants, and adult Carabid (Carabidae) and rove (Staphylinidae) beetles (Eisenbeis & Wichard 1985; Wallwork 1976). Fire Effects on Sandhill Litter and Soil Fauna The effects of fire on arthropods function through various mechanisms, at different times. Immediate effects of fire can be direct mortality (Miller 1955) forced emigration (Gillon 1970), and immigration of pyrophilous species attracted to the smoke or heat of fires. Short te rm effects after fires can include fireinduced changes to the hab itat which can strongly affect food qu ality and accessibility (Benzie 1986; Evans 1984; Rice 1932; Schowalter 2000), foraging sites (Andersen 1988, 1991), predation rates (Knutson & Campbell 1976; Schowalter 2000), microclimate (Andersen 1991; Samways 1990) and composition of predator and detritivore communities (Agee 1993; McCoy 1987). Longer term effects include changes in ecosystem processes, including primary production and nutrient cycling (Miller 1955).

PAGE 13

6 Often there are small or no long-term effects of fire on ordinal-level abundances of arthropods in temperate woodlands and open forests of south Australia (Andersen & Muller 2000), but fire management can sometimes have large effects on insects and their medi ated processes in North Am erica. One study that has found outbreaks of herbivorous ins ects resulting from fire suppression in North American forests also found that the outbreak, in turn, can lead to increased vulnerability of the forests (McCullough et al. 1998). Another study found that fire suppression resulted in a decrease in the abundance of ants while late fires resulted in a reduction in spider, homopteran and silv erfish abundances. Overall, most fauna were not found to be affected by fire (Andersen & Muller 2000). There has been some concern that re peated prescribed burns may be detr imental to invertebrate populations (Dietrich et al. 1998; Opler 1981; Reed 1997; Swengel 1994, 1996) but it seems that this may not always be the case. Although Hanula (2003) found the lowest diversity and evenness in the one-year fire frequency plots and highest in the control plots, Friend (1996) found that the overall composition of assemblages were often resilient to fire, even if substantial changes in arthropod abundances at the species level occurred. A greater number of hemiptera and homoptera were found on burned sites than unburned sites in one study (Cancelado & Yonke 1970). Higher biomass of herbivorous arthropods was also found on burned versus unburned sites (Nagel 1973). It has been found that fire intervals of 2 to 3 years do not affect species richness or composition. Most populations, especially those arthropod populations that are common in an area, have been found to recover within a growing season, although some may take up to 2 years or more (Hanula 2003; Panzer, 2002). Uncommon species recovery time may take longer or not occur at all (Bess et al. 2002). In a study of prescribed burning ef fects on spider and carabid beetles, it was found that the effects of burning can differ within taxonomic families. It was found that four out of seven spider families were more abundant on unburned sites than burned, while three of the seven families were more abundant on burned sites. It was also found that four of five beetle families were more abundant on unburned sites, while there was no difference found for the other family. Although abundances differed, species richness and diversity did not differ significantly for either spider s or carabid beetles (Niwa & Peck 2002). It does seem important for conservation of soil arthropods in fragmented habitats to provide for the various needs of the different arthropods in a habitat. It is also important to include different intensities and timing of prescribed burns and to include areas without burn s to allow for recruitment (Hanula 2003; Howe 1994; Panzer 2002; Swengel 1996). Perhaps the optimal fire frequency of 3 years found for many prairieinhabiting birds (Swengel 1996) could be compatible with that of arthropods on fragmented land as well (Panzer 2002). The dry sand of sandhill makes it a target for development. Fire suppression, logging, habitat destruction for development, as well as other human induced m odifications have decreased Longleaf pine/Turkey oak sandhill habitat to about 3% of its historic range. C onsidering Florida and Georgia, in acreage alone, it has been estimated that a 90% decrease in sandhill acreage has occurred in the last 60 years (Jensen & Carson 2001). Maintaining the existing sandhill habitat is essential to conserving this species-rich habitat. This is a difficult task because of the ever looming threat of past and presen t land usage and development and because of the necessity for regular low intensity fires (Jensen & Carson 2001). In this thesis I will address the question of whet her sandhill soil and litter-dwelling arthropod communities differ depending on prescribed fire frequency, time since fire, seasonal and/or amount of leaf litter. It is hypothesized that time since the last fire and amount of leaf litter will affect the arthropod communities more than fire frequency or season. The amount of time since the last fire is predicted to affect community composition the most because it affects canopy cover and amount of leaf litter which could affect the composition of the litter-dwelling arthropods.

PAGE 14

7 METHODS The study site was located within the University of South Florida Ecological Research Area (USF Eco Area figure 1), which is located in west-central Florida. It is composed of approximately 200 ha of pine flatwoods, riverine swamp, and Longleaf pine ( Pinus palustris )/turkey oak ( Quercus laevis ) sandhill. Sandhill occurs in approximately 20 ha of the USF Eco Area (Schmidt, 2005). Longleaf pine and slash pine ( Pinus elliotti ) dominate the overstory and turkey oaks dominat e the midstory, along with scattered cabbage palms ( Sabal palmetto). An understory is composed of mostly wiregrass, Aristida stricta forbs and saw palmetto ( Serenoa repens ) (Mushinsky, 1992; Mushinsky, 1985). About 12.4 ha of the sandhill habitat along the southern border of the USF Eco Area is divided into experimental burn plots. Since the mid1970s, the plots have been subjected to different prescribed fire regimes. Prescribed burns occur in oneyear, two-year, five-year and sevenyear intervals and there are two re plicates of each treatment. Two reference plots have not been burned since at least 1965 (Mushinsky, 1985). Plot sizes range from approximately 0.7 to 1.9 ha (Figure 2). Figure 1. Aerial photograph of the University of South Florida Ecological Research Area in northeast Tampa, FL. The experimental sandhill plots used in this study are outlined in yellow.

PAGE 15

8 Although the intended prescribed fire frequencies are in one-, two-, five -, and seven-year intervals, the actual fire frequencies have not always followed this pattern. Appendix A shows the actual fire frequencies in the USF Eco Area. Delays in burn dates occurred fo r some plots and there were no burns in some years. There were no burns in 1994, 1995, 1997, 2000, 2001, 2002, 2004 or 2005. The delays were associated with the lack of burn for 1997, 1999, 2000 and 2001. There was a one year delay for the two and sevenyear burn plots intended to be burned in 1997 and delays of 3 years for the one-year plots intended to be burned in 1999. The two-year and five-year plots intended to be burned in 2000 and 2001, respectively, were burned three and two years late, respectively. The last fire occurred in 2003, at which time the one-, two-, and five-year plots were burned. Experimental Design There were five regularly-spaced sample locations in each plot/treatment, approximately 20-30 m apart. The five sample locations were set up to allow for a representative sample of the different areas in each treatment, while trying to keep the distance between each sample far enough so that the sample locations would not influence the others. This resulted in five treatments: Reference, 1-year, 2-year, 5-year and 7year fire frequencies and two replicates: East and We st (Figure 2). There were a total of seven sampling occasions. Sampling occurred in September (leaf litter onl y), and November of 2004 and in January, April, August, November of 2005. Figure 2. Map of the burn plots. By Emily Minor {, 2000 #142}. Based on GPS points of the outline of the plots; the areas are calculated by ArcView based on those data In each treatment five small pitfall traps (16 oz plastic cups) were set in the ground with the top of each cup flush with the ground. Thes e cups were filled with about half an inch of antifreeze to preserve the specimens. The pitfall traps were shaded with 2 gallon bucket lids, propped up with sticks, and were left out for 5 consecutive days during each season (Figures 3 and 4). Between seasons, the pitfall traps were closed. The traps were checked daily when open. If rain occurred and the cups were more than 1/3 full of water, the contents were collected and the trap reset. If disturbance occurred, the traps were repaired and reset.

PAGE 16

9 Figure 3. The general set up of a pitfall trap. The tr aps used in this study contained antifreeze as the killing solution and did not have a hardware cloth mesh covering. Figure 4. The presentation of a pitfall trap. The pitfall traps used in this study were 16oz plastic cups, filled 1/8 full of antifreeze. A two gallon bucket lid, propped up by sticks was used to shield the trap contents from sun and rain.

PAGE 17

10 Leaf litter was sampled using a 625 cm2 quadrat made from PVC pipes, which allowed one or two plots to be sampled for leaf litter each day during the time that pitfall traps were checked for damage or inundation with rain water. The quadrat was th rown haphazardly near each of the fi ve sample areas per plot, and all leaf litter within the quadrat was collected. These samples were brought back to the lab on the same day they were collected. The mass and major vegetation type of each leaf litter sample were recorded for the April, August and November samples. To extract the arthropods from the leaf litter Tullgren funnel traps (Sutherland, 1996) were used. The traps had 100 watt bulbs that functioned to drive the arthropods out of the leaf litter and into the 70% ethanol preservative in the collecting jars below the funnel (Fig ures 5-7). The collecting jar of the funnel traps were composed of 16 ounce plastic cups filled with approxima tely one inch of ethanol. Leaf litter was left in the trap for 24 hours. The ethanol and preserved arthropods were poured into a vial and stored in a refrigerator. The samples were then presorted to extract the arth ropods from the alcohol and leaf litter debris. The arthropods were placed into a smalle r vial filled with ethanol (70%). Ar thropods were later identified into taxonomic order. (Quayle, 2001) Figure 5. Tullgren funnel apparatus. The apparatus used in this experiment consisted of a beaker and a plastic funnel with hardware cloth screen.

PAGE 18

11 Figure 6. Tullgren funnel set up for this experiment. Each 16 oz cup contained about an inch of 70% alcohol. Hardware cloth with mesh size 0.25 inch was used as a screen in the funnel. A 100 watt light bulb is placed directly over the funnel to drive insects down into the alcohol. Figure 7. Tullgren funnel trap station.

PAGE 19

12 Data Analysis Univariate Statistics Order richness (S) was calculated as the number of arthropod orders. Diversity indices were employed to measure diversity using the number of orders in the sample and the evenness of their distribution for the different treatments. Order abundances were calculated and used to calculate Shannon-Wiener coefficients of diversity (H’), Hills number 1 (N1)and Hills number 2 (N2). The Shannon-Wiener index H' = i pi log(pi) was used because variances can be calculated and ttests can be used to determine if the values are significantly different. Shannon-Wiener is also one of the diversity indices with the best discriminatory ability and ease of computation (Clarke and Warwick 2001). N1 is a transformation of the Shannon-Wiener diversity index: [exp(H’)] and N2 is the reciprocal of the Simpsons diversity index: [1/ ]. Hill’s evenness index (H) was used to calculate the evenness of relativ e abundances. Hill’s numbers were used because they are easier to interpret than other diversity index values (Hill 1973). A univariate ANOVA was used to test for treatment effect s of fire frequency, and time since the last fire on arthropod abundance, the number of arthropod orders present for all seasons pooled and for seasons separately. Shannon-Wiener index values were used to statistically analyze the data using a univariate ANOVA (Clarke and Warwic k 20011). ANOVA tests were performe d on arthropod abundances from each plot to test for the effect of fi re frequency on the arthro pod abundance. The different fire frequencies were the different levels of the burn factor. Although the data are not normal and do not have homogeneity of variance, post-hoc multiple comparison test that do not assume homogeneity of variances are not very efficient and do not have high power. ANOVA is robust to heterogeneity of variances and non-normality of data, so post-hoc multiple comparison tests that assume homogeneity of variances can be used (Zar 1999). If one uses an of 0.05 for an ANOVA and computes multiple t-tests, the probability of committing a type I error is increased for the entire experiment. Because of this, it is often invalid to perform multiple t-tests. Post-hoc tests use levels that are more conservative than 0.05. These are used so that, if the null hypothesis is rejected, the chance of committing a type I error is still 0.05 and not higher. If an ANOVA is found to be significant, Post-hoc tests are used to compare all groups (Madrigal 1998). There is no consensus as to which post-hoc procedure is the best to use but the Student-Newman-Keuls, Tukey’s and Tukey’s b tests are common methods. Tukey’s test is one of the more robust of the multiple comparison tests and it was used in my analyses (Zar 1999). A repeated measures ANOVA was used to test for differences within Shannon-Weiner diversity index values between sampling periods (Zar 1999). This would account for time effects and lack of independence of replicates over time. If there was a time effect, th is effect can be accounted for and then a test for treatment effect can be done. Spearman Rank Correlations were employed to test for correlations between arthropod abundance, number of arthropod orders, separately, and then leaf litter mass and fire frequency treatments and years since last fire, separately. The Spearman Rank correlation coefficient, rS, is an estimate rank correlation coefficient for the population and ranges from -1 to +1 (Zar 1999). Spearman Rank Correlations were employed to test for correlations between Shannon-Wiener diversity values and leaf litter mass treatments and leaf litter mass treatments and arthropod abundances. The Friedman two-way analysis of variance is simila r to a repeated measures ANOVA in that it was used to test for different community compositions among fire frequencies and time since the last fire (Zar 1999). The nonparametric Kendall’s coefficient of concordance (W) tests for agreement among rankings of all groups being considered. W values range from 0 for no agreement to 1 for full agreement among the ranking of all groups or complete (Zar 1999). The Wilcoxon signed-rank test was used to test th e symmetry of dispersion and is a nonparametric alternative to the paired Students t-test (Zar 1999).

PAGE 20

13 Multivariate Statistics Similarity between sampling units (traps, plots and fire frequency treatments) was measured with the BrayCurtis index (Bray and Curtis 1957) using the dispersion weighted, square-root transformed counts of order abundances to decrease the relative impor tance of overdispersed or highly clumped orders and to reduce the effects of very abundant orders, respectively (Clarke et al. 2006). Dispersion weighting is a procedure that addresses th e differences in the variance structure of individual order counts. Dispersion weighting downweights only the overdispersed or clumped orders, thereby, increasing the importance of rarer but consistent orders which have a more Poisson-like distribution. The clumped order counts were divided by their average index of dispersion D, which is the average variance to mean ratio for treatment groups, in order to get Poisson like values. Without changing the relative differences between mean responses within treatment groups, dispersion weighting results in a matrix with a Poisson-like variance structure across all orders. Dispersion weighting downweights counts with high variability, thereby causing a small increase in power wh ich can increase the significance in ANOSIM tests of group differences. The downweighting of highly variable counts is similar to increasing power in univariate statistics by decreasing error (Clarke et al. 2006). The order similarity matrix was generated from which later multivariate analyses in PRIMER (Primer-E) were performed. The unweighted group average method of cluster analysis was used to identify groups of traps, plots, and treatments with the most differen t community compositions. Post-hoc randomization was performed using the similarity profile test (SIMPROF) of PRIMER. SIMPROF performs 1000 random rearrangements of each variable, separately, across the samples to test for signifi cant differences in the composition of the community at each cluster dendogr am node (Clarke and Gorley 2006). For ordination, non-metric multi-dimensional scaling (MDS) was used as suggested by Ludwig and Reynolds (1998). The nonparametric analysis of similarities (ANOSIM) was performed to test for different community compositions among plots, fire frequencies and time since last fire treatments (Clarke et al. 2006). The RELATE routine of PRIMER was used to test for directional change in community composition by comparing the Bray Curtis similarity matrix to a co nstructed model matrix. Separate model matrices were constructed for burn frequency and the number of months since fire between all pairs of plots. Each model matrix was correlated with the Bray Curtis similarity matrix. Similarity equals one if two samples are completely similar and equals zero if the two samples have no orders in common. Similarities betw een replicates within a site are expected to be higher than the similarities between replicates at diff erent sites. Two orders are considered similar if they are represented at a significant amount of same sites and dissimilar if they never occur together (Clarke and Warwick 2001). The Bray-Curtis similarity coefficien t was used because of its common us e in ecology and because it is one of the few measurements that can satisfy the following conditions: when two samples have no orders in common it takes the value of zero, when an order is absent from the two samples and is included or excluded, the value is unchanged. The similarity be tween samples A and B are not changed if a third sample, C, is included or excluded in the data array and when the relative abundances for all orders are the same, it is elastic enough to show differences in total abundance for two samples as an imperfect similarity. Many of the other coefficients that also satisfy th ese criteria are either too similar to the Bray-Curtis coefficient or have drawbacks that make them undesirable (Clarke and Warwick 2001). Overall, the similarity measurement was utilized in or der to replace the original data set with a pairwise similarity matrix that will be easier to work with for further analyses. The similarities are chosen to reflect similarity in arthropod abundances in the community (Clarke and Warwick 2001). The similarity matrix was used to differentiate between sites and times. The si milarity indices were used because they take into account shared order abundances, ther eby decreasing any distortion caused by inadequately sampled rare species. Similarities within each group are exp ected to be higher than those between groups.

PAGE 21

14 The program PRIMER was used to measure the similari ties in order abundances or richness in different treatments (PRIMER-E 2006). This was used to measure the number of orders and individuals common to the communities compared. Similarities within and among treatment groups were calculated, using the Bray-Curtis coefficient and the absolute data values be tween every pair of the sa mples in the data matrix. Values range from 0-1, arranged in a similarity matrix. The similarity ma trix is a triangular arrangement of similarities between every pair of orders. It describes the pattern of occurrences of all orders in combination across the given set of samples (Clarke and Warwick 2001). The unweighted group average cluster method with a SIMPROF post-hoc randomization procedure was used to perform cluster analysis to group plots and treatments with similar communities. Clustering was used because the similarity matrix was too large to easily determine patterns of similarity. Replicate samples within a site that form a cluster that is distin ct from replicates within other sites can be considered to have different community compositions from the other sites. Sites or times can then be put into groups with similar community structure and order assemblages can be defined. (Clarke and Warwick 2001). The non-metric multi-dimensional scaling (MDS) ordination technique was performed to produce a graphical representation of the degree of similarity between sites (Clarke and Warwick 2001). Ordination techniques arrange data in relation to axes so that the relative positions of the data to each other, and to the axes, give the most information about their similarities. Ordination brings large data sets to simpler one-, two-, or S-dimensional figures that can be used to visualize similarity among communities (Krebs 1999). The relative placement of samples on the ordination map reflects the similarity of their communities. Distances between samples on the ordination axes are a reflection of the dissimilarities in the community structures (Clarke and Warwick 2001). To test for different community compositions among plots, fire frequency and time since last fire treatments, analysis of similarities (ANOSIM) was used. ANOSIM is analogous to the parametric analysis of variance procedure (ANO VA). ANOSIM was used because there ar e fewer assumptions to satisfy than there are for ANOVA. R is the test statistic for AN OSIM and it contrasts the differences among rank similarities between treatments with differences among replicates within treatments. R = 1 if all rank similarities within treatments are more similar than rank similarities between treatments. R = 0 if the null hypothesis of no treatment effects occurs. ANOSIM performs permutations of all possible arrangements of variable labels and calculates an R statistic for each arrangement. ANOSIM calculates an R statistic among all treatments, which is called the Global R. If the Glob al R is significant, R statistics are then calculated for each pairwise permuation of the treatments (Clarke and Warwick 2001). To test for directional change in community composition, the Bray Curtis similarity matrix was correlated with a constructed model matrix using the RELATE routine of PRIMER. A Monte Carlo randomization procedure calculates the Spearman rank correlation of the model matrix and Bray Curtis matrix and performs 999 permutations of random rearrangements of plot labels in the Bray Curtis matrix and calculates a statistic for each arrangement. (Clarke and Gorl ey, 2006). The test statistic for RELATE is the statistic. When is greater than 0.95, a very high degree of correlation between the model and Bray Curtis matrices exists (Clarke and Warwick 2001). Similar model matrices were constructed for fire frequen cy and months since the last fire between all trap pairs and the RELATE procedure was performed for their correlations to the Bray Curtix similarity matrix.

PAGE 22

15 RESULTS Overall A total of 15735 individuals from 14 orders, sampled over five occasions, were used in the analyses (Table 1). Arthropod abundance values ranged from 688 to 7689 individuals per sampling occasion (Table 1), and 2264 to 3554 for each fire frequency treatment across all sampling occasions (Tables 2-6). Collembola, hymenoptera and acariformes were the most abundant orders. Collembola had a peak abundance of 5559 in August 2005. Hymenopteran abundance peaked in April and August 2005 (1004 and 1126, respectively). Acariformes peaked in abundance in January 2005 with 681 individuals. These orders were the most abundant in the winter months as well. Spiders joined collembola and hymenoptera as dominators in spring, while collembola dominated in summer, followed by hymenoptera. Arthropod abundance was higher in the most frequently burned treatment and the unburned reference treatment (oneand over 30 year fire frequencies) than in the intermediate fire frequencies (two-, five-, and seven-year fire frequency) (Table 2). Arthropod abundances increased over time across all fire frequency treatments until August 2005, with one exception (the seven-year fire frequ ency in April 2005) (Table 5). Shannon-Weiner diversity was not different among fire frequency treatments (p: 0.091). The time since the last fire did affect Shannon-Weiner diversity during the January 2005 (p = 0.031), and April 2005 sampling occasion (p = 0.013). Shannon-Weiner diversity was not correlated with leaf litter mass treatments. The arthropod community distribution of plots was generally even (Table 3). Arthropod abundances in the November 2004 sampli ng occasion were positively correlated with fire frequency ( rS: 0.900, p: 0.031) and time since the last fire ( rS: 0.894, p: 0.041). In all other sampling occasions arthropod abundances were not correlated with fire frequency or time since the last fire. Arthropod abundances were not correlated with leaf litter mass treatments. The Friedman and Kendall’s W test revealed that arthropod abundances were different, overall, among sampling occasions across fire frequ encies (p: 0.001). Specifically, th e Wilcoxon test revealed that arthropod abundances were different for the November 2004 and January 2005, April and August 2005, August and November 2005, and for the November 200 4 and November 2005 samp ling occasions (p: 0.063 for each pair) across fire frequencies. The mean number of orders ranged from 10-12 orders for fire frequency treatments overall (Table 8) and varied for each sampling occasion (Table 9). The number of arthropod orders in any of the sampling occasion was not correlated with fire frequency. The number of arthr opod orders in the January 2005 sampling occas ion was positively correlated with time since the last fire ( rS: 0.894, p: 0.041). In all other sampling o ccasions the number of arthropod orders was not correlated with time since the last fire. The number of arthr opod orders was different, overall, among sampling occasions across fire frequencies (p: 0.043). The number of arthropod orders was different for the November 2004 and January 2005 sampling occasions (p: 0.077) and for the November 2004 and Novemb er 2005 sampling occasions (p: 0.042) across fire frequencies.

PAGE 23

16 Leaf litter mass in all sampling occasions where l eaf litter mass was recorded was positively correlated with fire frequency (Table 13). Leaf litter mass was also correlated with time since last fire for the April and August 2005 sampling occasions ( rS: 0.894, p: 0.041 for both). Leaf litter masses were different, overall, among all sa mpling occasions across fire frequencies (p: 0.074) and time since last fire (p: 0.074). Specifically, leaf litter masses were different for the April and November 2005 sampling occasions across fire fr equencies (p: 0.043) and time sin ce last fire (p: 0.043). Leaf litter masses from the August 2005 and November 2005 sampling occasions were different among time since last fire treatments (p: 0.003 and p: 0.013, respectively). Dispersion The dispersion weighting downweighted the highly variable orders. Dispersion weightin g corrected for the clumped distributions of arthropods. The data indicate that all orders have significantly aggregated patterns of dispersion (all D values are greater than 1 (Table 14)). D is the degree to which an individual order tends to be aggregated. The null hypothesis of random dispersion is supported if D = 1. The resulting dispersion weighted similarity matrix is found in Appendix B. Cluster analysis The cluster analyses of community composition among experimental plots indicated two major groups. One group is made up of plots with the more extreme fire frequencies: 1 year and 7 years between fires and unburned plots. The other group is made up of plots with more the intermediate fire frequencies of 2 and 5 years between fires. All twoand five-year plots were burned one year prior to data collection, as well as the one-year plots (SimProf at alpha = 0.05) (Figure 8). Cluster analysis dendrogram of experimental traps. Th e X-axis is labeled with traps. Solid lines indicate significant differences among plots at p = 0.05. Symbols indicate the amount of years since the last fire. There are two major groups. One group consists of the CWE trap, one of the traps that has not been burned for at least 30 years. The other group consists of all other traps with all the other times since last fire: 1 year, 7 year or at leas t 30 years (Figure 9). MDS The results of the nonmentric multidimensional scaling of experimental plots indicate significantly different groups. One group is made up of plots with fire frequencies of twoand five-years. The other group in made up of the other plots with fire frequencies of one-, seven-, and greater than 30-years.The goodness-of-fit is high, indicated by the stress measure of 0.05. Higher goodness of fit is indicated by lower stress (figure 10). The results of the nonmentric multidimensional scalin g of experimental traps indicate loose groups. The goodness-of-fit is relatively low, indicated by the stress measure of 0.24. Lower goodness of fit is indicated by higher stress (figure 11). ANOSIM The results of the two-way nested ANOSIM using traps as samples indicate no effect of plots on arthropod abundances (Global R = -0.042, p = 0.731; 999 permutations). Arthropod abundances in the fire frequency treatments were different when considering traps as samples (Global R = 0.67, p = 0.002; 945 permutations) which allowed a one-way ANOSIM to be performed with traps as samples. The results of the one-way ANOSIM indicate that ar thropod abundances were different among plots across all fire frequency treatments (Global R = 0.21, p = 0.002; 999 permutations) (Table 17). Community composition was different between unburned reference plots and all plots burned in November 2003, and between the seven year fire frequency and the two and five year treatments (Table 18).

PAGE 24

17 RELATE The results of the RELATE procedure, randomized Sp earman Rank correlation values of similarities among fire frequency treatments, indicate a correlatio n of Bray-Curtis similarities among plots and the pattern of differences among fire frequency treatments ( = 0.356, p = 0.021; 999 permutations). The results of the Relate procedure, randomized Spea rman Rank correlation values of similarities among years since the last fire, indicate a strong correlation of Bray-Curtis similarities among plots and the pattern of differences among years since fire ( : 0.492, p = 0.007; 999 permutations). Repeated Measures ANOVA Multivariate Tests indicate no significant effect of season on arthropod diversity (p = 0.121). Tests of within-subjects effects indicat e an effect of season on arthropod diversity for all criterion (p = 0.001, 0.008, 0.044) but no effect of the interaction of season and plot (p = 0.465, 0.459, 0.451) (Table 15). Post hoc tests did not indicate any difference among burn frequencies (p = 0.124). Seasonal Cluster analysis Without any a priori knowledge, the results of the cluster analysis that was performed for the seasonal data indicated that arthropod abundances from plots within similar sampling occasions were generally more similar to one another than to plots from differ ent sampling occasions. If another sampling occasion was found within a group, it wa s usually a sampling occasion adjacent to that of the group. Any outliers that were present were closest to the group of the same sampling occasion (Figure 12). MDS The results of the nonmetric multidimensional scaling indicate a general pattern of change over time, with the exception of the Novemb er 2005 sampling occasion. Again, samp les clumped together in groups from the same sampling occasion. Samples within the same sampling occasions also became more similar as time goes on (stress = 0.13) (Figure13). ANOSIM The results of the 2-way crossed ANOSIM indicate no difference among fire frequency groups across all sampling occasions (Global R = 0.048, p = 0.308; 999 permutations). Tests for differences among sample occasions indicated that sampling occasions were different across all fire frequency groups. (Global R = 0.538, p = 0.001; 999 permutations ). All sampling occasions were different except November 2004 and Jan 2005, and August 2005 and November 2005 (Table 16). Results of RELATE Spearman rank correlation The results of the Relate procedure, randomized Spea rman Rank correlation values of similarities among for time of year, indicate a correlation of Bray-Curtis similarities among plots and the pattern at each time of year.( : 0.078, p = 0.076; 999 permutations). The results of the Relate procedure, randomized Spea rman Rank correlation values of similarities among months since fire, indicate a correl ation of Bray-Curtis similarities am ong plots in each sampling occasion and the pattern of differences among months since fire ( : 0.078, p = 0.076; 999 permutations).

PAGE 25

18 Traps Table 1. 15735 individuals used in the analyses. The data from the September 2004 sampling occasion were not included in the analyses because pitfall traps were not employ ed during that sampling occasion. Number of Individuals Nov-04 688 Jan-05 1785 Apr-05 2167 Aug-05 7689 Nov-05 3406 Total 15735 Table 2. Total and Mean Arthropod Abundance per Fire Frequency. Fire Frequency Total Abundance Mean Abundance One Year 3554 710.8 Two Year 2727 545.4 Five Year 2264 452.8 Seven Year 3318 663.6 Reference – No Fire 3413 682.6 Table 3. Diversity, Richness and Evenness measures. S is the number of orders represented. N is the number of individual arthropods represented. H’ is the Shannon-Weiner diversity index, N1 and N2 are Hills Diversity indices, re spectively. E5 is the evenness index. J’ is the Jaccard evenness index. Plot S N J' H'(loge) 1-Lambda' N1 N2 E5 Fire Frequency (years) Years Since Fire 1E 14 2347 0.44 1.16 0.51 3.21 2.05 0.48 1 1 1W 14 1419 0.64 1.69 0.76 5.41 4.08 0.70 1 1 2E 13 1650 0.44 1.13 0.52 3.09 2.06 0.51 2 1 2W 14 1116 0.55 1.46 0.65 4.32 2.85 0.56 2 1 5E 13 1306 0.50 1.29 0.59 3.63 2.46 0.56 5 1 5W 14 1079 0.56 1.49 0.68 4.43 3.15 0.63 5 1 7E 13 2179 0.58 1.49 0.65 4.45 2.82 0.53 7 7 7W 14 1518 0.63 1.67 0.74 5.33 3.84 0.66 7 7 CE 14 1889 0.58 1.54 0.72 4.66 3.56 0.70 30 30 CW 13 2169 0.62 1.59 0.69 4.88 3.25 0.58 30 30 Table 4. Leaf litter mass is found to increase from Ap ril 2005 to August 2005 from means of 97.34g to 126.76g. The leaf litter mass between August 2005 and November 2005 were similar. Sampling Occasion S N J' H'(loge) 1-Lambda' N1 N2 E5 Total Leaf Litter Mass (g) Mean Leaf Litter Mass (g) Nov-04 13 627 0.52 1.34 0.65 3.81 2.88 0.67 N/A N/A Jan-05 14 2030 0.71 1.87 0.80 6.49 4.91 0.71 N/A N/A Apr-05 14 2093 0.62 1.64 0.72 5.16 3.54 0.61 486.71 97.34 Aug-05 14 7565 0.37 0.98 0.44 2.67 1.77 0.46 633.80 126.76 Nov-05 14 3318 0.49 1.29 0.53 3.63 2.14 0.43 616.06 123.21

PAGE 26

19 Table 5. Arthropod Abundances per Fire Frequency. Fire Frequency Nov-04 Jan-05 Apr-05 Aug-05 Nov-05 One Year 53 406 585 1841 669 Two Year 33 194 303 1373 824 Five Year 118 165 443 1151 387 Seven Year 166 502 330 1479 841 Reference – No Fire 257 406 432 1721 597 Table 6. Mean Arthropod Abundances per Time Since Last Fire. Time Since Last Fire Nov-04 Jan-05 Apr-05 Aug-05 Nov-05 Total One Year 68 255 444 1455 627 2849 Seven Years 166 502 330 1479 841 3318 Reference > 30 Years 257 406 432 1721 597 3413 Table 7. Mean Arthropod Order Richness per Fire Frequency. Fire Frequency Mean Order Richness One Year 12 Two Year 10 Five Year 10 Seven Year 12 Reference – No Fire 12 Table 8. Mean Order Richness per Sampling Occasion. Fire Frequency Nov-04 Jan-05 Apr-05 Aug-05 Nov-05 One Year 6 13 13 14 14 Two Year 5 10 12 12 13 Five Year 9 8 13 10 12 Seven Year 9 14 13 13 13 Reference – No Fire 7 14 12 13 13

PAGE 27

20 Table 9. Total and Mean Leaf Litter Mass for each Plot and Sampling Occasion. Sampling Occasion Plot Total (g) Mean(g) Apr-05 1E 273 54.6 1W 200 40.0 2E 300 60.0 2W 204 40.8 5E 325 65.0 5W 657 131.4 7E 634 158.5 7W 627 125.4 CE 437 87.4 CW 1070 214.0 Apr-05 Overall 4727 472.7 Aug-05 1E 386 77.2 1W 419 83.8 2E 300 60.0 2W 400 80.0 5E 446 89.2 5W 530 106.0 7E 786 157.2 7W 432 86.4 CE 996 199.2 CW 1643 328.6 Aug-05 Overall 6338 633.8 Nov-05 1E 371 74.2 1W 444 88.8 2E 467 93.4 2W 463 92.6 5E 644 128.8 5W 350 70.0 7E 1151 230.2 7W 725 145.0 CE 645 129.0 CW 746 186.5 Nov-05 Overall 6006 600.6 Table 10. Mean Leaf Litter Mass per Fire Frequency. Fire Frequency Mean Leaf Litter Mass (g) One Year 69.77 Two Year 71.13 Five Year 98.40 Seven Year 149.84 Reference – No Fire 189.72

PAGE 28

21 Table 11. Mean Leaf Litter Mass per Fire Frequency for each Sampling Occasion. Sampling Occasion Fire Frequency Mean Leaf Litter Mass (g) Apr-05 1 Year 47.30 2 Year 50.40 5 Year 98.20 7 Year 140.11 Reference No Fire 150.70 Aug-05 1 Year 80.50 2 Year 70.00 5 Year 97.60 7 Year 121.80 Reference No Fire 263.90 Nov-05 1 Year 81.50 2 Year 93.00 5 Year 99.40 7 Year 187.60 Reference No Fire 154.56 Table 12. Mean Leaf Litter Mass per Time Since Last Fire. Time Since Last Fire Mean Leaf Litter Mass (g) One Year 79.77 Seven Years 149.84 Reference – No Fire 189.72 Table 13. Spearman Rank correlations for leaf litter mass and fire frequency for sampling occasions where leaf litter mass was recorded. Spearman Rank Correlations for Leaf Litter rS P Apr-05 1.0 > 0.001 Aug-05 0.9 0.037 Nov-05 0.9 0.37

PAGE 29

22 Table 14. Dispersion weighting is used to downweight highly variable orders. The null hypothesis of random dispersion is supported if D = 1. Index of Dispersion (D) Coefficients D Sig% Divisor Collembola 109.176 0.000 109.176 Acariformes 23.183 3.381 E-187 23.183 Spiders 1.987 0.009 1.987 Formicidae 11.908 3.148 E-83 11.908 Hymenoptera 5.004 8.450 E-24 5.004 Diptera 3.492 2.414 E-12 3.492 Coleoptera 4.977 1.375 E-23 4.977 Homoptera 1.450 1.700 1.450 Hemiptera 1.959 0.000 1.959 Thysanoptera 9.632 5.830 E-63 9.632 Psocoptera 5.099 0.000 5.099 Isoptera 2.963 0.000 2.963 Blattodea 2.159 0.000 2.159 Orthoptera 1.998 0.008 1.998 Null = dispersion is randon (D = 1). Significant: not random (% >5 = not significant; p>0,05

PAGE 30

23 Plots Figure 8. Cluster analysis dendrogram of arthropod co mmunity composition among experimental plots. Labels indicate plot. Solid lines indicate significant differences among plots at p = 0.05. There are two major groups. One group consists of plots with the more extreme fire frequencies: 1 year and 7 years between fires and unburned plots. The other group consists of plots with more intermediate fire frequencies of 2 and 5 years between fi res. All twoand five-year plots were burned one year prior to data collection, as well as the one-year plots.

PAGE 31

24 Figure 9. Cluster analysis dendrogram of experimental trap s. X-axis is labeled with traps. Solid lines indicate significant differences among plots at p = 0.05. Symbols indicate the amount of years since the last fire. There are two major groups. One group consists of the CWE trap, one of the traps that has not been burned for at least 30 years. The ot her group consists of all other traps with all the other times since last fire: 1 year, 7 year or at least 30 years.

PAGE 32

25 Figure 10. Plot of nonmetric multidimensional scaling of experimental plots. Plot designations are labeled with colors and shapes. Dark blue triangles refer to pl ots with1 year fire frequencie s, teal squares refer to plots with 2 year fire frequencies, red diamonds refer to plots with 5 year fire frequencies, purple circles refer to plots with 7 year fire frequencies, and gray pl us signs refer to the reference plots without a fire for at least 30 years. Contour lines in dicate significantly different groups (SimProf at alpha = 0.05). The goodness-of-fit is high, indicated by the stress measure of 0.05. Higher goodness of fit is indicated by lower stress.

PAGE 33

26 Traps Figure 11. Plot of nonmentric multidimensional scaling of experimental traps. Trap designations are labeled with colors and shapes. Dark blue triangles refer to pl ots that were burned 1 year prior to data collection, teal squares refer to plots that had not been burned for 7 years prior to data collection, and red diamonds refer to plots that have not had a fire for at least 30 years prior to data collec tion. The goodness-of-fit is relatively low, indicated by the stress measure of 0.24. Lower goodness of fit is indicated by higher stress.

PAGE 34

27 Table 15. Repeated measures analysis of variance (MANOVA) for differences within fire frequencies between sampling periods. Tests of Within-Subject s Effects indicate signifi cant effects of season. Source Measure Type III Sum of Squares df Mean Square F Sig. Noncent. Parameter Observed Power(a) Season Sphericity Assumed 1.891 4 0.473 7.161 0.001 28.643 0.98 GreenhouseGeisser 1.891 2.333 0.811 7.161 0.008 16.703 0.88 Huynh-Feldt 1.891 4 0.473 7.161 0.001 28.643 0.98 Lower-bound 1.891 1 1.891 7.161 0.044 7.161 0.577 Season DumPlot Sphericity Assumed 1.111 16 0.069 1.052 0.451 16.834 0.448 GreenhouseGeisser 1.111 9.33 0.119 1.052 0.459 9.817 0.3 Huynh-Feldt 1.111 16 0.069 1.052 0.451 16.834 0.448 Lower-bound 1.111 4 0.278 1.052 0.465 4.209 0.166 Error (Season) Sphericity Assumed 1.32 20 0.066 GreenhouseGeisser 1.32 11.663 0.113 Huynh-Feldt 1.32 20 0.066 Lower-bound 1.32 5 0.264 a Computed using alpha = .05

PAGE 35

28 Seasonal Figure 12. Cluster analysis dendrogram of plots for eac h sampling occasion. X-axis is labeled with plots for each sampling occasions. Solid lines in dicate significant differences among plots at p = 0.05. Plots seem to clump with other plots of the same sampling occasion with the exception of a few plots which are usually from an adjacent sampling occasion. Any outliers that are present are closest to the group of the same sampling occasion.

PAGE 36

29 MDS The results of the nonmetric multidimensional scalin g of sampling occasions indicate seasonal groupings that correspond with arthropod abundance. Figure 13. Plot of nonmetric multidimensional scaling of sampling occasions. Sampling occasion designations are labeled with colors and shapes. Th e goodness-of-fit is relatively good, indicated by the stress measure of 0.13. Higher goodness of fit is indicated by lower stress. Two-Way Crossed ANOSIM: Seasonal Table 16 Pairwise tests of sampling occasions across all fire frequency groups. Groups R Statistic Significance Level % Possible Permutations Actual Permutations #>=Obs Nov04, Jan05 0.4 0.095 243 243 23 Nov04, Apr05 0.6 0.016 243 243 4 Nov04, Aug05 0.95 0.004 243 243 1 Nov04, Nov05 0.7 0.004 243 243 1 Jan05, Apr05 0.6 0.012 243 243 3 Jan05, Aug05 0.8 0.004 243 243 1 Jan05, Nov05 0.65 0.016 243 243 4 Apr05, Aug05 0.6 0.008 243 243 2 Apr05, Nov05 0.5 0.025 243 243 6 Aug05, Nov05 0.4 0.074 243 243 18 7565 3318 2093 1673 627 A rthro p od Abundance

PAGE 37

30 One Way ANOSIM: Seasonal Table 17. Pairwise tests of sampling occasions. Groups R Statistic Sig Level % Possible Permut Actual Permut #>=Obs Nov04, Jan05 0.392 0.001 92378 999 0 Nov04, Apr05 0.598 0.001 92378 999 0 Nov04, Aug05 0.793 0.001 92378 999 0 Nov04, Nov05 0.693 0.001 92378 999 0 Jan05, Apr05 0.409 0.001 92378 999 0 Jan05, Aug05 0.752 0.001 92378 999 0 Jan05, Nov05 0.514 0.001 92378 999 0 Apr05, Aug05 0.755 0.001 92378 999 0 Apr05, Nov05 0.455 0.001 92378 999 0 Aug05, Nov05 0.546 0.001 92378 999 0 Table 18. Pairwise tests of fire frequency and arthropod abundances. Groups R Statistic Sig Level % Possible Permut Actual Permut #>=Obs 30, 1 0.15 2.5 92378 999 24 30, 2 0.412 0.1 92378 999 0 30, 5 0.587 0.1 92378 999 0 30, 7 0.074 9.8 92378 999 97 1, 2 -0.003 45.6 92378 999 455 1, 5 0.083 9.5 92378 999 94 1, 7 0.072 11.5 92378 999 114 2, 5 -0.054 76.4 92378 999 763 2, 7 0.318 0.3 92378 999 2 5, 7 0.455 0.1 92378 999 0

PAGE 38

31 DISCUSSION Fire determined the number of arthropods and th e community composition without affecting the speciesabundance distribution. When all sampling occasions were pooled, arthropo d abundance and ShannonWiener diversity were not affected by fire frequency. Th is result has been found in other studies as well. In an inland mixed conifer forest, although abundances differed, species richness and diversity did not differ significantly for either spiders or carabid beetle s (Niwa & Peck 2002). Community composition in my study, however, was affected by fire frequency. This is because orthopterans were more abundant in plots that were burned in November 2003, while all other orders of arthropods were more abundant in plots that were not burned recently. Orthopterans increase in abundance shortly after fire (Kerstyn & Stiling 1999; Knight & Holt 2005). Fire increases biomass of herbaceous vegetation (Gli tzenstein et al. 2003) and quality of forage (Radho-Toly et al. 2001). The experimental plots were relatively small, which may have allowed many arthropods that respond positively to fire to m ove into areas with higher fire frequencies and those that respond negatively to move into areas with lower fire frequencies. Arthropod abundance and diversity were affected by season. At different times of the year arthropod abundances and diversity were different. Certain groups of orders peaked at different times of the year. Orders that peaked in summer and fall (late in the vegetative growing season) included collembolans, orthopterans, coleopterans and dipterans. Many of these orders feed on vegetation and their peaks in abundance follow peaks in new vegetative growth. Orde rs that peaked in the dormant season (late fall through spring) included isopterans, psocopterans, mites and thysanopterans. Many species in these groups feed on dead plant matter which is more abundant in the dormant season. Two groups peaked in abundance during the early growing season: hymenopterans and spiders. Many hymenopterans and all spiders are predaceous and their peaks in abundance may be tracking peaks in abundance of their prey. It is also likely that many species in these orders may be responding to annual variation in cl imate, such as summer rainfall, which can vary significantly from one year to the next at the study site (Halstead 2007). The effects of fire on arthropods function through various mechanisms that occur at different times. Immediate effects of fire can be direct mortality (Miller 1955), forced emigration (Gillon 1970), and immigration of pyrophilous species attracted to the smoke or heat of fires. Short term effects after fires can include fire-induced changes to the habitat which ca n strongly affect food quality and accessibility (Benzie 1986; Evans 1984; Rice 1932; Schowalter 2000), foraging sites (Andersen 1988, 1991), predation rates (Knutson & Campbell 1976; Schowalter 2000), microclimate (Andersen 1991; Samways 1990) and composition of predator and detritivore communities (Agee 1993; McCoy 1987). Because orders peak in abundance at certain times of the year, the season in which plots are burned may have different effects on arthropod communities. Orders and life stages that are dispersal-limited may be negatively affected if fires occur wh en they are near their peak abundan ce or at their lowest abundances. A dormant season fire may affect some orders more than others, and vice versa. Seas on of fire affects plant communities. Dormant season fires can allow hardwoods to persist at greater distances (Myers 1990). Changes in plant species composition associated with timing of fire are also very likely to affect community compositions of arthropods utilizing these plant species. Arthropod responses to disturbance can vary. Lepidopt eran, detritivore and predator abundances decreased while sap-sucking insect abundances increased in response to canopy-opening disturbances such as hurricanes and timber harvesting (Schowalter 1994; 1995). Grasshopper (Kerstyn & Stiling 1999; Nagel 1973), leaf miner (Kerstyn & Stiling 1999) and herp etofaunal (Mushinsky & Gibson 1991; Halstead 2007) abundances were found to be higher in the more frequently burned plots at this study site. Friend (1996) found that the overall composition of assemblages were ofte n resilient to fire, even if substantial changes in arthropod abundances at the species level occurred.

PAGE 39

32 Differing time since the last fire, which is known to affect species diversity, co uld be attributed to the mixed results found (Gibson & Hulbert 1987; Mehlman 1992). It has been found that fire intervals of 2 to 3 years do not affect species richness or compositio n. Most populations, especially those arthropod populations that are common in an area, have been found to recover within a growing season, although some may take up to 2 years or more (Hanula 2003; Panzer, 2002). Uncommon species recovery time may take longer or not occur at all (Bess et al. 2002). W ith the exception of mites, every order increased in abundance from November 2004 to November 2005. This may suggest that post-fire recolonization may still have been occurring. As noted earlier, however, th ese increases in abundance may be a result of annual environmental variation. Leaf litter mass increased with increasing time since the last fire, but was not correlated with diversity. Perhaps time since the last fire has an indirect effect on diversity. The amount of time since the last fire affects leaf litter mass, which then affects diversity. In January 2005, it had been 14 months since the last fire. This meant that leaf litter was allowed to accumul ate during that time. In April, even more time had passed, allowing more leaf litter to accumulate. Perhaps this affected the arthropod community. It is not known why diversity in November 2004 was not affected by the amount of time since the last fire. Because the November 2004 sampling occasion was clos est to the most recent burn date, one might expect differences in diversity to be st ronger and lessen as plots became mo re similar in successional stage. However, it is also possible that because different arthropods are active at diff erent times of the year, arthropod community composition may be more sensitive to the amount of leaf litter in January and April than in August and November. Studies have been inconsistent in determining the relationship between fire and diversity. The relationship between arthropod abundance, diversity and community composition and fire could be too subtle to detect with the experimental design used, or there could have been low power attributed to the relatively small number of replicated treatments and the short sampling period of the study. Although not possible in the site where the research was conducted, additional replication beyond that of east and west would be a way to increase sample size. Sampling only occurred over a 12 month period. A longer period of sampling would have increased the sample size and decreased any variation attributed to sampling. For reasons beyond my control, the prescribed fire that was intended for 2005 never occurr ed, therefore, immediate responses after a fire were not able to be observed. If there had been a longer sampling period, perhaps sampling would have occurred just before and after the next prescribed fire to detect more immediate changes in the arthropod communities. Data collection also didn ’t begin until a year after the most recent fire. Immediate effects of fire are probably strongest and they were probably missed. In a study of prescribed burning ef fects on spider and carabid beetles, it was found that the effects of burning can differ within taxonomic families (Niwa & Peck 2002). The level of taxonomic classification needed to understand the patterns of a community can differ among organisms (Hirst 2006). Although not possible in this study because of lack of funding, identifying arthropods to a finer taxonomic resolution than order may have been more effective for understanding the patterns of diversity, arthropod abundance and community composition in the sandhill community that was studied. Organisms within taxonomic families such as coleoptera differ so much in their en vironmental requirements, and tolerances that genus level identification may be necessary for this analysis. It is likely, given the large number of species within orders and the broad range of ecological responses of individual species, that species-specific responses may be hidden by the broad taxonomic level. It does seem important for conservation of soil arthropods in fragmented habitats for land managers to provide for the various needs of the different arthropods in a habitat. It is also important to include different intensities and timing of prescribed burns and to in clude areas without burns to allow for recruitment (Hanula 2003; Howe 1994; Panzer 2002; Swengel 1996). Perhaps the optimal fire frequency of 3 years found for many prairie-inhabiting birds (Swengel 1996) could be compatible with that of arthropods on fragmented land as well (Panzer 2002).

PAGE 40

33 Arthropods in the sandhill habitat of the USF ecological research areas seem to respond more strongly to seasonality than to fire frequency. Fire frequency may be important but likely it may be more important at certain types of year or to certain groups of arthropods. A lot of the response to fire may have been masked by the low taxonomic resolution, low replication, short sampling period and length of time passed between the last fire and the beginning of sampling. Further study should include a longer sampling period including periods immediately before and after a fire. This would allow for observation of seasonal effects and immediate effects of fire. It would also allow for separation of the time since the last fire and fire frequency treatments, something that was unable to be done in this study because all the recently burned plots were also the most frequent burn treatments. Analysis of pitfall and leaf litter traps separately wo uld also be helpful because they seemed to target different compositions of arthropods. Arthropods should also be identified to at least the family level and maybe even the genus level if possible. There is almost as much diversity in several of the larg er families of beetles as is found in the entirety of most of the other arthropod orders (Arnett 2000). To conserve time and resources, perhaps beetles could be used as a subset of the arthropods present to test for differences at a higher taxonomic resolution. Replication which includes other areas of sandhill ha bitat would also help to increase the power of the study. If these changes were included in the next stud y, perhaps the response of sandhill soil and leaf litter dwelling arthropods to fire may be better understood. A quote by Friend (1996) seems to sum up this study most eloquently, “invertebrate populations are so variable in time and space, being larg ely driven by environmental factors, that the impacts of fire may be largely over-ridden and almost impossible to predict. This is particularly the case where a broad level of taxonomic resolution is adopted.”

PAGE 41

34 LITERATURE CITED Abrahamson, W. G. (1984a). Post-fire recovery of Florida Lake Wales Ridge vegetation. Am. J. Bot. 71:921. Abrahamson, W. G. (1984b). Species responses to fire on the Florida Lake Wales Ridge. Am. J. Bot. 71:3543. Agee, J. K. (1993). Fire Ecology of Pacific Nort hwest Forests. Island Press, Washington, D.C. Ahlgren, I. F. (1974). The effects of fire on soil organisms in T. T. Kozlowski, and C. E. Ahlgren, editors. Fire and Ecosystems. Academic Press, New York. Andersen, A. N. (1988). Immediate and longer-term effects of fire on seed predation by ants in sclerophyllous vegetation in south-eastern Australia. Aust. J. Ecol. 13:285-293. Andersen, A. N. (1991). Responses of ground-foraging ant communities to three experimental fire regimes in a savanna forest of tropical Australia. Biotropica 23:575-585. Andersen, A. N., and W. J. Muller. (2000). Arthropo d responses to experimental fire regimes in an Australian tropical savannah: ordinal-level analysis. Austral. Ecology 25:199-209. Beckage, B., and I. J. Stout. (2000 ). Effects of repeated burning on species richness in a Florida pine savanna: A test of the intermediate disturbance hypothesis. Journal of Vegetation Science 11:113-122. Benzie, J. A. H. (1986). The distribution, abundance, and the effects of fire on mound building termites ( Trinervitermes and Cubitermes spp., Isoptera: Termitidae) in nort hern guinea savanna, West Africa. Oecologia 70:559-567. Bess, E. C., R. R. Parmenter, S. McCoy, and J. M.C. Molles. (2002). Responses of a Riparian Forest-Floor Arthropod Community to Wildfire in the Middle Rio Grande Valley, New Mexico. Environmental Entomology 3:774-784. Bradstock, R. A., and P. J. Myerscough. (1981). Fi re effects of seed release and the emergence and establishment of seedlings n Banksia ericifolia L.F. Aust. J. Bot. 29:521-531. Bray, J. R., and J. T. Curtis. (1957). An ordination of the upland forest communities of southern Wisconsin. Ecological Monographs 27:325-349. Brockway, D. G., K. W. Outcalt, and R. N. Wilkins. (1998). Restoring longleaf pine wiregrass ecosystems: plant cover, diversity and biomass following low-ra te hexaxinone application on Florida sandhills. Forest Ecology and Management 103:159-175. Buchholz, K. (1983). Initial responses of pine and oak to wildfire in the New Jersey Pine Barren Plains. Bulletin of the Torrey Botanical Club 110:91-96. Cancelado, R., and T. R. Yonke. (1970). Effect of prairie burning on insect populations. Journal of the Kansas Entomological Society 43:274-281. Cheplick, G. P., and J. A. Quinn. (1987). Effect of fire on community structure in tall and mixed-grass prairie. Oecologia 73:459-464.

PAGE 42

35 Chust, G., J. L. Preus, D. Ducrot, A. Bedos, and L. Deharveng. (2003). Identification of landscape units from an insect perspective. Ecography 26:257-268. Clarke, K. R., Chapman M.G., Somerfield P.J., Needham H.R. (2006). Dispersion-based weighting of species counts in assemblages analys es. Mar Ecol Prog Ser 320:11-27. Clarke, K. R., and R. N. Gorley (2006). PRIMER v6: User Manual/Tutorial. PRIMER-E Ltd, Plymouth. Clarke, K. R., and R. M. Warwick (2001). Change in Marine Communities: An Approach to Statistical Analysis and Interpretation. PRIMER-E Ltd, Plymouth. Collins, S. L., S. M. Glenn, and D. J. Gibson. (1995). Experimental analysis of intermediate disturbance and initial floristic composition: decoupling cause and effect. Ecology 76:486-492. Daubenmire, R. F. (1968). Ecology of fire in grasslands. Adv. Ecol. Res. 5:203-226. Dietrich, C. H., M. G. Harper, R. L. Larimore, and P. A. Tessene. (1998). Insect s and fire: too much of a good thing? Illinois Natural History Reports 349:5. Dodd, C. K., and B. G. Charest. (1988). The herpetofaunal community of temporary ponds in north Florida sandhills: species composition, temporal use and management implications. Pages 87-97. In symposium proceedings: Management of Amphibi ans, Reptiles, and Small Mammals in North America. U.S. Dept. of Agriculture, Forest Servi ce, Rocky Mountain Forest and Range Experiment Station, Fort Collins. Eisenbeis, G., and W. Wichard (1985). Atlas on the biology of soil arthropods. Springer-Verlag, Berlin. Evans, E. W. (1984). Fire as a natural disturbance to grasshopper assemblages of tallgrass prairie. Oikos 43:9-16. Force, D. (1981). Postfire insect succession in southern California chaparral. American Naturalist 117:575582. Friend, G. (1996). Fire ecology of invertebrates implications for nature conservation, fire management and future research. Pages 155-162. Proceedings of Fire and Biodiversity. The effects and effectiveness of fire management. Biodiversity Unit, Department of the Environment, Sport and Territories, Canberra, Australia. Gibson, D. J., and L. C. Hulbert. (1987). Effects of fire, topography and year-to-year climate variation on species composition in tallgrass prairie. Vegetatio 72:175-185. Gill, A. M. (1977). Plant traits adaptive to fires in Mediterranean -type ecosystems. Pages 17-26 in H. A. Mooney, and C. E. Conrad, editors. Symposium on the Environmental Consequences of Fire and Fuel Management in Mediterranean Ecosystems. USDA. Gillon, D. (1970). Recherches ecologiques dans la savane de Lamto (Cote d'Ivoire): les effets du feu sur les Arthropodes de la savane. Terre Vie 24:80-93. Glitzenstein, J. S., D. R. Streng, and D. D. Wade 2003. Fire frequency effects on longleaf pine ( Pinus palustris P. Miller) vegetation in South Carolina and no rtheast Florida, USA. Natural Areas Journal 23: 22-37. Gullan, P. J., and P. S. Cranston (2000). The Ins ects: An Outline of Entomology. Blackwell Science Ltd., Oxford.

PAGE 43

36 Hale, W. G. (1976). Collembola. Pages 397-411 in A. Burges, and F. Raw, editors. Soil Biology. Academic Press, London. Halstead, N.T. (2007). Long-term ef fects of prescribed fire on reptile and amphibian communities in Florida sandhill. M.S. Thesis, Univer sity of South Florida, Tampa, FL. Hansen, R. A. (2000). Effects of habitat complex ity and composition on a diverse litter microarthropod assemblage. Ecology 81:1120-1132. Hanula, J. L. (2003). Influence of long-term dormant-season burning and fire exclusion on ground-dwelling arthropod populations in longleaf pine flatwoods ecosystems. Forest Ecology and Management 175:163-184. Hartnett, D. C. (1987). Effect of fire on clonal growth and dynamics of Pityopsis graminifolia (Asteraceae). Am. J. Bot. 74:1737-1745. Hay-Smith, L., and G. Tanner. (1997 a). Responses of Wildlife Food Plants to a Forest Herbicide. Department of Wildlife Ecology and Conservation: Range Science Program. University of Florida, Gainesville. Hay-Smith, L., and G. Tanner. (1997b). Restoring Long leaf Pine Sandhill Communities with an Herbicide. Department of Wildlife Ecology and Conservation: Range Science Program. University of Florida, Gainesville. Hill, M. O. (1973). Diversity and evenness: A unifying notation and its consequences. Ecology 54:427-432. Hirst, A. J. (2006). Influence of taxonomic resolution on multivariate analyses of arthropod and macroalgal reef assemblages. Mar Ecol Prog Ser 324:83-93. Hobbs, R. J., J. E. P. Currall, and C. H. Gimingham. (1984). The use of 'thermocolor' pyrometers in the study of health fire behavior. J. Ecol. 72:241-250. Hobbs, R. J., and C. H. Gimingham. (1984). Studies on fire in Scottish heathl and communities. I. Fire characteristics. J. Ecol. 72:223-240. Howe, H. F. (1994). Managing species diversity in tallgrass prairies: assumptions and implications. Conservation Biology 8:691-704. Hughes, J. B., G. C. Daily, and P. R. Ehrlich. (2000). Conservation of Insect Diversity: a Habitat Approach. Conservation Biology 14:1788-1797. Humphrey, J. W., C. Hawes, A. J. Pearce, R. Ferris-Khan, and M. R. Jukes. (1999). Relationships between insect diversity and habitat complexity in plantation forests. Forest Ecology and Management 113:1121. Huston, M. A. (1994). Biological Diversity: The coexis tence of species on changing landscapes. Cambridge University Press, Cambridge. Jensen, R. R., and A. R. Carson. (2001). The Changing South: Longleaf Pine/Turkey Oak Sandhill loss in a North Central Florida preserve, 1972-1997. Southeastern Geographer 41:306-311. Kalisz, P. J., Dorian A.W., Stone E. L. (1986). Prehisto ric land-use and the distribution of longleaf pine on the Ocala National Forest, Florida: an interdisciplinary synthesis. Fla. Anthropol. 39:183-193. Keeley, J. E. (1986). Resilience of Mediterranean shrub communities to fire Pages 95-112 in B. Dell, A. J. M. Hopkins, and B. B. Lamont, editors. Resilience in Mediterranean Ecosystems. W. Junk, Dordrecht.

PAGE 44

37 Kerstyn, A., and P. Stiling. (1999). The effects of bur n frequency on the density of some grasshoppers and leaf miners in a Florida sandhill community. Florida Entomologist 82:499-505. Knapp, A. K., and L. C. Hulbert. (1986). Production, density and height of flower stalks of three grasses in annually burned and unburned eastern Kansas tallgrass prairie. Southwest. Nat. 31:235-242. Knutson, H., and J. B. Campbell. (1976). Relationships of grasshoppers (Acrididae) to burning, grazing, and range sites of native tallgrass prairie in Ka nsas. Proc. Annu. Tall Timbers Conf. Ecol. Anim. Control. Habitat Manage. Komarek, E. V. (1974). Effects of fire on temperat e forests and related ecosystems: southeastern United States. Pages 251-277 in A. C. E. a. T. T. Kozlowski, editor. Fire and Ecosystems. Academic Press, New York. Krebs, C. (1999). Ecological Methodology. Addison Wesley Longman, Inc, Menlo Park. Liberton, C. Undated. Will sandhill have a home in Hernando County? Hernando Chapter News. Hernando Chapter of the Florida Native Plant Society, Brooksville. Longino, J. T., and R. K. Colwell. (1997). Biodiversity assessment using structured inventory: capturing the ant fauna of a tropical rain forest. Ecological Applications 7:1263-1277. Ludwig, J. A., and J. F. Reynolds (1988). Statistical Ecology: A primer on methods and computing. John Wiley & Sons, New York. Madrigal, L. (1998). Statistics for Anthropology Cambridge University Press, Cambridge, UK. Malanson, G. P., and L. Trabaud. (1987). Ordina tion analysis of components of resilience of Quercus coccifera garrigue. Ecology 68:463-472. McCoy, E. D. (1987). The ground-dwelling beetles of periodically-burned plots of sandhill. Florida Entomologist 70:31-39. McCullough, D. G., R. A. Werner, and D. Neumann. (1998). Fire and insects in northern and boreal forest ecosystems of North America. Annual Review of Entomology 43:107-127. Mehlman, D. W. (1992). Effects of fire on plant community composition of North Florida second growth pineland. Bulletin of the Torrey Botanical Club 119:376-383. Miller, R. B. (1955). Biological and chemical changes following scrub burning on a New Zealand hill soil. NZ J. Sci. Technology 37:290-313. Mushinsky, H. R. (1992). Natural history and abundances of Southeastern Five-lined Skinks, Eumeces inexpectatus on a periodically burned sandhill in Florida. Herpetologica 48:307-312. Mushinsky, H. R., and D. J. Gibson. (1991). The influence of fire periodicity on habitat structure Pages 237-259 in E. D. M. S.S. Bell, H.R. Mushinsky, ed itor. Habitat Structure: the physical arrangement of objects in space. Chapman & Hall, London. Mushinsky, H. R. (1985). Fire and the Florida sandh ill herpetofaunal community: with special attention to responses of Cnemidophorus sexlineatus Herpetologica 41:333-342. Myers, R. L. (1990). Scrub and High Pine. Pages 150-193 in R. L. Meyers, and J. J. Ewel, editors. Ecosystems of Florida. University of Central Florida Press, Orlando. Nagel, H. G. (1973). Effect of spring prairie burning on herbivorous and non-herbivorous arthropod populations. Journal of Kansas Entomological Society 46:485-496.

PAGE 45

38 Niwa, C. G., and R. W. Peck. (2002). Influence of pres cribed fire on carabid beetle (Carabidae) and spider (Araneae) assemblages in forest litter in southwes tern Oregon. Environmental Entomology 31:785796. Nuzzo, V. A., W. McClain, and T. Strole. (1996). Fire impact on groundlayer flora in a sand forest. Am. Midl. Nat. 136:207-221. Opler, P. A. (1981). Management of prairie habitats for insect conservation. Natural Areas Journal:3-6. Panzer, R. (2002). Compatibility of Prescribed Burning with the Conservation of Insects in Small, Isolated Prairie Reserves. Conservation Biology 16:1296-1307. Platt, W. J., G. W. Evans, and S. L. Rathbun. (1988). Effects of fire season on flowering of forbs and shrubs in longleaf pine forests. Oecologia 76:353-363. PRIMER-E, L. (2006). PRIMER v6. PRIMER-E Ltd, Plymouth. Quayle, J. (2001). Berlese (Tullgren) funnels. Province of British Columbia, http://ilmbwww.gov.bc.ca/risc/pubs/tebiodiv/terranth/arthml20-2-03.htm. Radho-Toly, S., J. D. Majer, and C. Yates. 2001. Impact of fire on leaf nutrients, arthropod fauna and herbivory of native and exotic eucalypts in King s Park, Perth, Western Au stralia. Austral Ecology 26:500-506. Raw, F. (1967). Arthropoda (except Acari and Collembola). Pages 323-362 in A. Burges, and F. Raw, editors. Soil Biology. A cademic Press, London. Reed, C. C. (1997). Responses of prairie insects and other arthropods to prescr iption burns. Natural Areas Journal 17:380-385. Rice, L. A. (1932). Effects of fire on the prairie animal communities. Ecology 13:392-401. Samways, M. J. (1990). Land forms and winter habitat refugia in the conservation of montane grasshoppers in southern Africa. Conservation Biology 4:375-382. Samways, M. J. (2005). Insect Diversity Conservation. Cambridge University Press, Cambridge. Schmidt, A. C. (2005). A Vascular Plant Inventory and Description of the Twelve Plant Community Types Found in the University of South Florida Ecological Research Area. Page 118. Department of Biology. University of South Florida, Tampa. Schowalter, T. D. (1994). Invertebrate community structure and herbivory in a tropical rainforest canopy in Puerto Rico following Hurricane Hugo. Biotropica 26:312-319. Schowalter, T. D. (1995). Canopy arthropod communities in relation to forest age and alternative harvest practices in western Oregon. Forest Ecology and Management 78:115-125. Schowalter, T. D. (2000). Insect Ecology: an ecosystem approach. Academic Press, San Diego. Schwilk, D. W., J. E. Keeley, and W. J. Bond. (1997). The intermediate disturbance hypothesis does not explain fire and density patterns in fynbos. Plant Ecol. 132:77-84. Siemann, E., J. Haarstad, and D. Tilman. (1997). Short-term and long-term effects of burning on oak Savanna arthropods. Am. Midl. Nat. 137:349-361. Sutherland, W. J. (1996). Ecological Census Techniques: a handbook. Cambridge University Press, Cambridge.

PAGE 46

39 Swengel, A. B. (1994). Skippers over the prairie: Da kota and Ottoe skippers fl y free. American Butterflies 2:4-9. Swengel, A. B. (1996). Effects of fire and hay management on butterfly abundance in tallgrass prairie and pine barrens. Biological Conservation 76:73-85. Tansley, A. G. (1924). The classification of vegetation and the concept of development. J. Ecol. 8:118-149. Tester, J. R. (1996). Effect of fire frequency on oak savanna in east-central Minnesota. Bulletin of the Torrey Botanical Club 116:134-144. Uetz, G. W. (1979). The influence of variation in litter habitats on spider communities. Oecologia 40:29-42. Wallwork, J. A. (1967). Acari. Pages 363-395 in A. Burges, and F. Raw, editors. Soil Biology. Academic Press, London. Wallwork, J. A. (1976). The Distribution and Diversity of Soil Fauna. Academic Press, London. Ware, S., C. C. Frost, and P. Doerr. (1989). Southern mixed hardwood forest (the former longleaf pine forest). in W. H. Martin, editor. Biotic Communities of the Southeastern United States. Wiley, New York. Warren, S. D., D. J. Scrifes, and P. D. Teel. (1987). Response of grassland arthropods to burning: a review. Agric. Ecosyst. Environ. 19:105-130. Wein, R. W., and L. C. Bliss. (1973). Changes in arctic Eriophorum tussock communities following fire. Ecology 54:845-853. Zar, J. H. (1999). Biostatistical Analysis. Prentice Hall, Upper Saddle River.

PAGE 47

40 APPENDICES

PAGE 48

41 APPENDIX A: DATES OF PRESCRIBED BURNS AT THE USF ECO AREA Before 1976, burns were inconsistent and may have been natural fires. Data were obtained by inspection of maps by Bruce Williamson. Check marks indicate intended prescribed burn schedule. Date of fire indicates when the actual fire occurred. 1968 2W burned 4/4 1971 2W burned 5/4 1976 5E, 5W, 7W burned 1/15 1979 1E, 2E burned 5/3 1E 1W 2E 2W 5E 5W 7E 7W Date of Fire 1979 5/3 1980 5/29 1981 6/10 1982 5/15 1983 5/27 1984 5/29 1985 5/16 1986 5/27 1987 6/25 1988 6/15 1989 6/16 1990 7/12 1991 7/18 1992 6/30 1993 7/20 1994 None 1995 None 1996 8/2 1997 None 1998 8/20 – 2 and 7 yr burned a year late 1999 8/28 2000 None 2001 None 2002 None 2003 10/27 – 1 yr plots burned 3 years late & 11/24 – 2 & 5 yr plots burned 3 and 2 years late, respectively 2004 None 2005 None Revised from (Schmidt, 2005). Compiled September 1998, Updated March 2006

PAGE 49

42 APPENDIX B: DISPERSION WEIGHTED AND SQUARE ROOT TRANSFORMED SIMILARITY MATRIX Bray-Curtis Similarity Between Plots CE CW 1E 1W 2E 2W 5E 5W 7E 7W CE CW 88.43 1E 87.89 80.72 1W 89.36 83.57 88.35 2E 80.81 74.39 85.93 84.04 2W 83.97 75.54 86.40 86.19 90.10 5E 79.24 73.16 86.27 82.27 89.99 87.78 5W 79.75 70.56 87.53 83.07 89.90 89.44 90.01 7E 88.31 90.61 85.20 88.00 79.33 79.02 76.22 74.98 7W 87.79 86.38 85.20 89.59 80.90 83.13 82.54 79.99 91.03