Savatzlttx 3:13.414... 1.. 1 I \...V.;: I .a I b .1: - 15.11.1251..I.;a:alill.:::)b& n 3.51 .31 f!!!§1f£i;:$§lbl)l'§)x {53:1 A‘§I.el.l:r.lutlfi:l)tl51.11.31.er gaiiign|1> at. 3.9;); .35.}; (Krill—vii. A a; ai§l$i§i¥§3§>~cfl .. .s 13‘ #IIISI , , . . . , , . . {.Eaxi I are}: .1, . g (1..-.-. :12..-$:Z I. . :3. z ‘3 nil! ~ All .1. . m . 5.1.; .51. .2‘ 1: 1:13.71 .3 3. 333.... 5... 13.1213: :t::r..r..1..a .35 .iaJ. I \ waif-:3 a... .92.}. 27:53. 23.... x1 .3. :ihrx‘x‘vsxx. .31....3‘147. nu. .Iula is I z l3ylv .ra‘xt .1... : u.vl..\..,rl. ,WIX .zv. (u... f... . . 253...: .251. . x fill? lllll"Illlllllllllllllllllllllllllllllllllllllllllllllll 193 01563 5208 This is to certify that the thesis entitled ACTIVITY AND DIVERSITY 0F CARABID GROUND BEETLES (COLEOPTERA: CARABIDAE) FOLLOHING PRESCRIBED BURNING IN A NATURE RED AND HHITE PINE STAND IN SOUTHERN MICHIGAN presented by David Donald Hermann has been accepted towards fulfillment of the requirements for Masters degree in FORESTRY Major professor Date May 5, 1997 0-7639 MSU is an Affirmative Action/Equal Opportunity Institution LIBRARY Michigan State Unlverelty PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. DATE DUE DATE DUE DATE DUE l MSU Is An Affirmative Action/Equal Opportunity Institution chWnfl-nt ACTIVITY AND DIVERSITY OF CARABID GROUND BEETLES (COLEOPTERA: CARABIDAE) FOLLOWING PRESCRIBED BURNING IN A MATURE RED AND WHITE PINE STAND IN SOUTHERN MICHIGAN By David Donald Neumann A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Forestry 1 997 ABSTRACT ACTIVITY AND DIVERSITY OF CARABID GROUND BEETLES (COLEOPTERA: CARABIDAE) FOLLOWING PRESCRIBED BURNING IN A MATURE RED AND WHITE PINE STAND IN SOUTHERN MICHIGAN By David Donald Neumann The effects of prescribed burning at 2- and 6-year intervals on Carabid beetle (Coleoptera: Carabidae) populations were investigated in a mature pine stand using pitfall trapping techniques during 1994 and 1995. Hypotheses relating beetle activity to frequency of prescribed burning and vegetation characteristics were tested. Total beetle activity was greater on burned than unburned areas. Differences were significant in 1994 between 6-year burn and unburned plots, but were not in 1995. Carabid species diversity was greater in both years on burned treatments than on unburned plots, as measured by the Shannon, Berger-Parker, Maraglef, Shannon evenness, log normal (it), and log series (on) indices. Beetle diversity patterns fit the log-series model most Closely. I conclude that Carabid diversity was not inversely related to fire frequency in this study. Two-year, 6-year, and unburned plots differed significantly in seedling and sapling density, while percent cover of ground flora differed only between burned and unburned plots. Regression analysis relating Carabid beetle trap catches to vegetation data did not significantly explain beetle activity. To Marylin Jean Neumann, 9-4-42 to 3-5-96. ACKNOWLEDGMENTS This document is long-winded enough; I would like to thank foremost those who helped keep it to the point. Without the guidance and encouragement Of Dr. Donald I. Dickmann, who helped more than most by simply having faith in this study, I would not have completed this work. Dr. Deborah McCullough and Dr. Karen Potter-Witter also provided invaluable inspiration, advice, and criticism during throughout this project. Dr. Mel Koelling has been an indispensable advisor, friend, and employer throughout my 9 years in this institution. Special thanks go to Dr. Robert Acciavetti of the USDA USFS in Morgantown, WV for verifying and correcting my initial type collection. I also thank Drs. Renata and Richard Snyder for their kind loan of collection jars, and initial instruction on beetle trapping. Many others deserve my thanks, including Sarah E. Synowiec, Dr. Daniel Keathley, David and Brad Boilore, the MSU Entomology Museum, Tim Work, and the MSU Department of Forestry. This work was supported in part by an award from the McIntire-Stennis Program. Finally, would like to thank my father, Robert W. Neumann, who taught me how to work hard, to be ambitious, and, as he is wont to yell while pounding on his desk, to JUST DO IT! TABLE OF CONTENTS LIST OF TABLES LIST OF FIGURES Chapter 1-- Introduction Effects of Fire on Vegetation Effects of Fire on Insects Goals and Research Hypothesis Chapter 2 -- Methods Experiment Site and Design Beetle Sorting Vegetation Sampling and Weather Statistical Analysis Species Diversity Analysis Chapter 3 -- Results and Discussion Overstory and Woody Understory Ground Flora Weather Effects vii ix 1O 15 16 17 23 26 37 47 Chapter 3 -- (continued) Carabid Beetle Relative Activity Beetle Categorical Data Analysis Vegetation Characteristics of Beetle Activity Species Diversity Qualitative Aspects of Species Diversity Chapter 4 -- Conclusions and Summary List of References Vi 52 58 65 75 85 88 92 LIST OF TABLES TABLE Page 1. Number of beetles in count data Classes used to categorize the relative activity of seven Carabid beetle species for logistic regression analysis. 22 2. Sampling dates used for logistic regression analysis for each of the seven most active Carabid species. 22 3. Average overstory basal area (mzlha) among burning treatments. 27 4. Seedling (s 1.9 cm at dbh) and sapling {2.0-5.9 cm dbh size class) density per ha in burned and unburned treatments. 29 5. Seedling species composition of 6-year burn and unburned control plots as the percent of total seedling density within each treatment ( 31.9 cm dbh size class). 31 6. Mean seedling ( 31.9 cm dbh) species density (per ha) in burned and unburned plots. 32 7. Species listed in “Other” category (present at <5% of treatment density) in Tables 7 and 8. 33 8. Sapling species composition of 6-year burn and unburned control plots as the percent of total seedling density within each treatment (2-5.9 cm size Class). 35 9. Mean sapling (2.0-5.9 cm dbh) species density (per ha) in burned and unburned plots. 36 10. 1994 woody ground flora species mean percent coverage by burn interval. 38 Vii TABLE Page 11. 1994 herbaceous ground flora species mean percent coverage by burn interval. 39 12. 1995 woody ground flora species mean percent coverage by burn interval. 40 13. 1995 herbaceous ground flora mean percent coverage by burn interval. 41 14. Ranked means tests for 1994 herbaceous and woody ground flora percent coverage. 43 15. Ranked means tests for 1994 herbaceous and woody ground flora percent coverage. 44 16. Number of ground flora species by year and burn interval. 44 17. Total number of Carabid beetles trapped by species and burn treatment. 55 18. Mean and total number of Carabids trapped by burn treatment in the nine most active species during 1994 and 1995 seasons. 56 19. Logistic regression analysis results from trap counts of the seven most commonly trapped Carabid species, and total Carabid catch in 1994 and 1995 trapping seasons. Probability values ( Pr > x2 ) and Odds ratios are listed for independent variables from categorical data analysis. 59 20. Block totals for nine common Carabid species and total number Of beetles trapped in 1994 and 1995. 61 21. Summary of significant variables in logistic regression models relating vegetation characteristics to beetle activity. 66 22. Diversity indices by year for burned and unburned treatments. 76 23. T-tests1 for Shannon diversity index estimates by year and burning treatment. 79 24. Goodness-Of-fit summary (or = 0.05) for four common species abundance models. 79 viii Figure 1. 2. LIST OF FIGURES Page Location of the experiment site. 11 Experiment layout map. 12 Demarcation of sampling periods into seasons based on similar rates of degree day accretion and accumulation in 1994 and 1995 (base 50 F). Trapping periods were 48 hours and occurred bi- weekly May 14-Oct. 16 in 1994 and May 12-Oct. 13 in 1995. 24 Average and minimum temperatures during trapping seasons for 1994 and 1995. Each point represents the average temperatures for the three days that comprise the trapping period, from opening to trap Closure. 49 Degree day accretion and accumulation during the trapping season in 1994 and 1995 (base 50 F). 50 Cumulative and 2-week incident precipitation over the trapping seasons in 1994 and 1995. Each point represents the average of precipitation occurring during the two-week period preceding and including trapping dates. 51 Total number of Carabid beetles trapped during each trapping period in 1994. 53 Total number of Carabid beetles trapped during each trapping period in 1995 54 Chapter 1 INTRODUCTION I keep six honest serving men (They taught me all I knew): Their names are What and Why and When and How and Where and Who. -Rudyard Kipling, ‘The Elephant’s Child” Pine trees and fire have long been natural, synergetic forces shaping northern forest ecosystems. Periodic fires swept through pre-settlement pine forests at intervals of 15 to 50 years in the Lake States region (Burgess and Methven 1977, Bergeron and Brisson 1990, Engstrom and Mann 1991, Maissurow 1941, Pyne 1982, Rouse 1988, Van Wagner 1970). Catfaces can still be seen on the boles of trees in old growth pine forests, attesting to their tolerance of periodic fire (Dickmann 1993). Red pine (Pinus resinosa Alt.) and white pine (Pinus strobus L.) are two commercially important pine species common to the Lake States region that reproduce vigorously on fire-prepared seed beds and appear resistant to periodic surface fires in mature stands. The natural predisposition of these species to coexist with fire leads to consideration of prescribed underburning as a management option in ecosystems in which they dominate (Dickmann 1993). While demands on woodlands for forest products mount, private and public managers have been pressed to manage 2 forest resources from an ecosystem perspective. Managers must also guide forest product procurement activities toward sustainable rates. Controlled underburning has been suggested as an environmentally benign way of solving some of the problems associated with intensive culture of mature pine stands. Several studies have described the effects of periodic low intensity prescribed fire on mixed red and white pine stands in Michigan (Dickmann, et al. 1987, Henning 1992, Henning and Dickmann 1996). Dickmann (1993) has published a thorough discussion of the multiple use benefits of prescribed burning in pine stands. Yet before prescribed burning can be widely applied, we need to better understand the effects of fire on all components of these ecosystems, including the multitude of invertebrates that inhabit them. The effects of wildfire or prescribed burning on terrestrial forest arthropods, especially in northern boreal forests, are still poorly understood. The preponderance of research on forest arthropod responses to disturbance in North America has focused on timber harvesting, although some studies have compared fauna communities in stands of different successional stages (Liebherr and Mahar 1979). Relatively few researchers have investigated changes in terrestrial arthropod fauna following fire, despite the importance Of fire as a natural disturbance regime in forested regions and as a management tool. Most fire- related research on insects has focused on changes in soil and litter layer inhabiting micro-fauna following prescribed burning or low to moderate intensity wildfires. (Metz and Farrier 1971, Buffington 1967; Heyward and Tissot 1936; 3 Ahlgren 1974; Rickard 1969; Moya-Raygoza 1995; Pearse 1943; Johnson 1995; Anderson et al. 1989; Jorgensen and Hodges 1970; Metz and Dindal 1975). Fire effects on boreal forest Carabid communities has been investigated by only two studies in North America. Holliday (1984), and Richardson and Holliday (1982) investigated Carabid community changes in burned spruce forests in Canada. Viereck and Dyrness (1979) recorded changes in soil arthropod communities following fire in Alaska, briefly noting the response of Carabids. Other studies of Carabid responses to prescribed burning have been completed in more temperate communities (Harris and Whitcomb 1974; Greenberg and Thomas 1995) or grasslands. Inferences from these studies may not be safely extrapolated to Lake States forest ecosystems. Effects of Fire on Vegetation Red and white pine overstories tolerate light understory fires very well, due to the excellent insulating qualities of pine bark (Riefsnyder et al. 1967). This quality is augmented by vigorous resin production that helps stems of these species to recover from moderate cambial wounds (Heinselman 1981 ). Underburning does not adversely affect overstory tree growth, provided that crown scorch is limited (Alban 1977, Methven and Murray 1974). Red pine can tolerate up to 75% crown damage from fire before dying (Van Wagner 1965, 1970). Lunt observed that 20 years of annual underburning actually increased height and volume growth in red pine stands in Ontario (1950). 4 Fire has been effective in reducing undesirable hardwood undergrowth in pine stands (Dickmann et al. 1987, Henning 1992, Henning and Dickmann 1996). On better soils or following thinning, red and white pine stands quickly develop a thick community of hardwood species. This proliferation becomes a boon if conversion to hardwood species following final overstory harvest is prescribed; a curse if the pine overstory is to be regenerated. Obtaining adequate regeneration on sites with undesirable hardwood competition is the major problem in red pine management in the Lake States region (Methven and Murray 1974). Periodic underburning kills the emergent portions of most hardwood undergrowth, encouraging low coppice growth in species capable of sprouting. Reduced competition for light also allows herbaceous vegetation to become re-established. Prescribed burning appears to affect herbaceous species composition more than percent coverage. Red maple (Acer rubrum), grasses , and bracken fern (Pten‘dium aqui/inum) are common plants under wide-spaced red pine in Michigan (Henning 1992). Burning encourages disturbance-favored species like raspberry (Rubus spp.), which reproduce vigorously after fire by sprouting (Ahlgren 1979). During the first three growing seasons following a single burn, Vigorous growth of herbs and shrubs can be expected (Ahlgren 1976). Herbaceous vegetation increased in percent coverage with increasing frequency of prescribed burning on sandy sites in New Jersey (Buell and Cantlon 1953). Frequent burning caused the development of a comparatively lush undergrowth of herbs and grasses. Fire stimulates perennial herbs, grasses and forbs to 5 root-sucker and sprout vigorously. Burning can also stimulate flowering of monocots (Trabaud 1987), and increase the nutritional quality of browse and forage in pine stands for wildlife (Dickmann 1993). Effects of Fire on Insects The effect of fire on forest invertebrates varies with burn intensity, community type, season, and region. Environmental changes following fire may be more important in determining population dynamics than the direct effects of heat and smoke (Lyon et al. 1978). Micro-Climatic and vegetation changes following fire may render conditions unsuitable for surviving species by removing shelter, sources for herbivory and prey. Following a moderate intensity fire in the pine barrens of New Jersey, Buffington (1967) Observed a decrease in arthropod abundance caused by a loss of incorporated and unincorporated soil organic matter that served as a food source for soil micro- and mesofauna. Hayward and Tissot (1936) attributed decreases in fauna inhabiting the Ao horizon of burned sites to post-fire xeric soil conditions. Changes in micro- habitat and vegetation structure may expose invertebrates to increased risks of predation (Anderson et al. 1989), while host plant response has been shown to have important effects on the abundance of some species. Several researchers have noted the effects of elevated soil temperatures and grass growth following prairie fires on grasshopper species abundance (Anderson et al. 1964 & 1989, Evans 1984, 1988, Rice 1932, Hurst 1971). Species mobility plays an important role in determining the temporal extent of fire effects. Metz and Farrier (1973) 6 found a 70 percent reduction in the number of acarians inhabiting the upper 7 cm of soil following fire. Populations recovered within 43 months, however. Anderson et al. (1989) noted significantly lower abundance of insect species on sand prairie sites only during the first year after prescribed burning. Rice (1932) found that many species declined in abundance during the fall after burning in prairie lands, but recovered with the spring flush Of vegetation.’ Some species are attracted and favored by fire disturbance in forests. Bark beetles and wood borers cause considerable damage and economic loss in stands stressed by fire. Pests that home in on trees injured or scorched by fire include the mountain pine beetle (Dendroctonus ponderosae), the pine engraver (Ips pin:), the western pine beetle (Dendroctonus brevicomis), the red turpentine beetle (D. valens), the Douglas fir tussock moth (Orgyia psuedotsugata), and the pandora moth (Coloradia pandon’a). Low-intensity prescribed burning can sometimes be used to help manage these populations, by destroying litter and woody debris where insects over-winter (Mitchell and Martin 1980, Miller 1978, Simmons et al. 1977). Frequency of fire disturbance has been shown to be an important factor in determining invertebrate population responses. Grasshoppers occurring on annually burned sites appear to be well adapted to climatic and vegetation changes associated with burning (Evans 1984). Gillon (1972) observed that populations of sun-loving Pentatomidae were maintained by periodic fires in African savannas. Riechert and Reeder (1972) suggest that spider populations 7 in dry prairie sites in Wisconsin are well adapted to fire, noting that periodically burned areas supported stable populations. The effect of fire on insect abundance and richness has been. studied by several researchers using Carabid beetles. Carabid beetle assemblages are convenient for studying the processes of recovery from fire because they are widely abundant, and have a well developed taxonomic nomenclature. Carabids are generally reduced in abundance for a short period of time following fire. Holliday (1991) observed that most Carabid species become locally extinct during intense forest fires or the extreme habitat changes that follow. Other researchers noted a 60 percent decrease in beetle populations in southern pine stands (Pearse 1943, Heyward and Tissot 1936). Ahlgren (1974) found fewer beetles in prescribed burned jack pine stands in Minnesota. Rapid colonization by pyrophilous and disturbance-tolerant species occurs following fire (Evans 1984); however, occupation by some species may be transient, lasting less than 2 years in boreal forests (Holliday 1984). Long- term population levels may remain depressed for 10 or more years in some forest stands following intense fires. Fire effects on Carabid beetle assemblages in grasslands may be less severe, partly due to lower fire temperatures. Tester and Marshal (1961) noted increases in both biomass and abundance of beetles on prescribed-burned power line rights-of-way. The effects of fire on Carabid species diversity are varied. Harris and Whitcomb (1974) noted that annually burned southern pine plantation sites a supported lower numbers of individuals and species than sites that had been protected for 10 years. Holliday (1992) noted no differences between species diversity in mature stands and burned sites in two boreal forest types over a 10- year period following wildfire. By contrast, Winter (1980) noted greater numbers of species and individuals in young pine stands for 1 to 3 years after burning in Germany. The reactions of Carabid fauna to disturbance may reflect the magnitude of the disturbance. During colonization of mining spoils by forests, researchers noted a weak trend toward increasing Carabid species abundance and diversity with increasing successional age in one study in Europe (Neumann 1971 ), but not in two others (Mader 1986, Hejkal 1985). Increased diversity Of Carabid species assemblages on clear-cut relative to uncut forest sites has been noted by several researchers (Niemala et al. 1993, 1993, 1990, Jennings et al. 1986, Lenski 1982). Important mechanisms governing insect diversity responses to disturbance may include intraspecific competition, size of the potential colonist pool for given sites, and the rate of Change in plant species diversity following fire. Holliday (1992) reported estimates of species gain and loss for boreal forest sites in Manitoba on burned and unburned sites following an intense forest fire. Unfortunately, little is known regarding Carabid species colonization and extirpation rates in temperate forest regions. Goals and Research Hypothesis The focus for the research project reported in this thesis was to investigate the effects Of prescribed burning at different frequencies on Carabid beetle species population diversity and abundance or activity. The study also examined the relationship between beetle observations and vegetation characteristics. The following hypotheses were tested: Global Hypothesis. Carabid beetle activity and species diversity in pine plantations are governed indirectly by vegetative responses to fire and not directly by fire frequency. Corollary Hypotheses. 1. Total Carabid beetle activity increases with the frequency of fire disturbance and diminishes with time since last fire. 2. Total Carabid activity is negatively correlated with density of the woody understory but unrelated to ground (herbaceous) vegetation composition or ground coverage. 3. Carabid species diversity (number and evenness of species) decreases with the frequency of fire and increases with time since the last fire within the two- year burn treatments. Chapter Two METHODS Experiment Site and Design The study was conducted at the Kellogg Experimental Forest, located in Kalamazoo County in southwest Lower Michigan, T1 S R9W (Figure 1). This experiment was established within an on-going prescribed burn study in Compartment 7 of the forest, a 4-ha. mixed plantation Of red and white pine. The stand was established in 1932 on an eroding hillside, and had been variably thinned over the years. The site occupies well drained sandy loam soils Of the Kalamazoo and Oshtemo series, derived of glacial ti" and outwash parent material. The Oshtemo soils are classified as coarse-loamy, mixed, mesic Typic Hapludalfs. Kalamazoo series soils are classified as fine-loamy, mixed, mesic Typic Hapludalfs. Site index varied with slope position, but was higher in block 1 on the Kalamazoo soils. The burn study was established in 1991 to study the ecological responses to surface fires ignited at different intervals. The plantation was divided into nine plots of about 0.4 ha, and one of three treatments assigned to each: unburned (control), burned every two years, and burned every six years (Figure 2). 10 l j // // /\ll 12 Figure 2. Experiment layout map. Mature Red Pine (Compt. 250) 'rafl 2!!!! ,. W4 m. I“. Mai 'm ‘V'1:§_ 5.1.33 ;.Cm1 -II 8, .’:’!£i" ‘2'. 1' TI'IY‘E‘IBI 4' "' ‘2 .1:" :LfifiAm °°°° Samar: rmauwgfigf gm. ”51:1“:le Iardwob‘! : 23:13:. i;i::::;::'::::: :‘I:',"~'.F‘ ::::: I! $!=tiil""':l!*2'1tltl g“;;g::':;:'::. d5} : Ii'fi‘tttfif '12::;::‘:::::': :‘,;.fi"",: :31! ‘ """ "“ ".i ‘.'!?I-!-5!-5i1-I-L slau- Ii .. ‘7 "“5:i'i'i'i:i:i""“1" 3"“ ::::: ‘‘‘‘‘‘‘ coo- ...... 000000 ooooooo eeeeeee ....... Control- 2-year 6-year iififtfigmmmm «I,» in mg .0 (It (A) 3" m ..... 0-4ha ....... ....... oooooooo OOOOOOOO ........ ooooooo ...... IIIIII ....... 0000000 oooooo ...... ....... ooooooo ....... nnnnnnn ooooooo °°""°" 6-year ...... CCCCCCC ooooooo ooooooo ------- nnnnnnnn cccccccc -------- oooooooo ........ oooooooo ....... ooooooo ooooooo ccccccc ooooooo ooooooo nnnnnnnn ....... IIIIIIII ooooooo nnnnn ..... ....... Line 56!“? ‘2-year 323:; Right-of-Way =55? (Brush) 0.46 ha 04 h *5; . a 0.45 ha OOOOOOOO ------- IIIIIII IIIIIII ooooooo ....... ooooooo ....... ooooooo ....... 0000000 ooooooo oooooo ..... ..... Mja_in Road I I Mature White Pine - Hardwood (Compt. 4B) N4 Scale: 1 cm = 15.5 m 13 Treatments were arranged in a randomized complete block design with three replications. Replicates were assigned to blocks based on slope position (upper, mid, lower). Plots in the two-year-burn treatment had been burned twice with low intensity surface fires at the beginning of my study (May of 1991 and 1993), and were re-burned in May of 1995. Six-year-burn plots had been burned once, in 1991. Fires generally burned within prescription, with flame lengths < 1.5 m and little scorching damage to the crowns of overstory trees, but almost complete topkill of the woody understory (Dickmann, pers. comm. 1995). To begin to understand invertebrate responses to the prescribed burn treatments, Carabid beetle populations were sampled during 1994 and 1995 using pitfall traps. The use of pitfall traps in population studies has been criticized by researchers (Adis, 1979; Greenslade, 1964) chiefly Citing their dependence on population densities and species activity characteristics. Trap catches are biased by the degree to which ground vegetation impedes Carabid movement. Behavior of different species can also reduce the effectiveness of pitfall traps, depending on position of the trap. Traps buried with the rim flush with the mineral soil surface tend to be most effective in trapping terrestrial and litter dwelling species; least effective in collecting arborescent species. Briggs (1961) and Greenslade (1964) deemed the use Of pitfall trapping catch data for assessing relative activity of Carabid species within the same habitat acceptable for strictly locomotor activity, however. Niemela et al. (1986, 1990b) recommend use of pitfall trapping mainly for use in comparing relative abundance within species, and in assessing habitat effects on Carabid diversity. 14 Other researchers found pitfall traps useful for monitoring populations (Luff, 1975) and for assessing relative population levels between habitat types (Baars, 1979; Reeves et al., 1983). Traps were systematically located on a 15-m grid pattern within each treatment plot. Six traps were lOcated within each plot, for a total of 54. Most trap locations were > 2 m from plot boundaries, however, three traps were within 1 to 2 m of plot edges. Trapping intervals were a compromise between a desire to collect the maximum number of species present on sites, and concerns that the trapping without replacement protocol might deplete local populations. Traps were activated biweekly for a 48-hour period from late May to mid October each year, beginning on May 14 in 1994, and on May 12 in 1995. Sampling in 1994 corresponded to the second post-fire year on biennial burn treatments, and the third year on 6-year-burn plots. Biennial burn treatments were re-burned in May of 1995. Therefore, by sampling the two-year-burn plots during 1995, a complete rotation of this treatment was encompassed. Trap catches were not compared between years, due to concern that the data might reflect two distinctly different causal events. In other words, the burns in the 2-year interval plots in 1993 and in 1995 may not be considered fixed effects. Pitfall traps consisted of an 8-cm diameter plastic cup (470 ml) and four, 1-m long, 10cm tall galvanized steel barriers arranged radially at 90° intervals around the collection cup. The cups were buried with the top of the container level with the mineral soil surface. Trapping cups were then covered and left undisturbed for one week prior to activation to avoid the “digging in” phenomena, 15 or temporarily elevated insect activity caused by increased CO; levels in the soil following disturbance (Joosse and Kapteijn, 1968). Trap arrays were left in place over the 1994-1995 winter. Traps in the two-year-burn treatment were temporarily removed prior to the burn in May of 1995. Most were returned to their original locations. To correct for spotty burning of the spring 1995 fire, seven traps were relocated to completely burned micro sites within 3 m Of their original locations. Traps were charged with ethylene glycol as a killing solution and preservative. Ethylene glycol was used throughout the 1994 trapping season, and no traps were disturbed during trapping periods. In an effort to mollify'public concerns about the toxicity of ethylene glycol, a more benign dilute solution of water mixed with non-scented, clear detergent was employed as a killing agent during the first three trapping dates in 1995. This protocol resulted in the loss of 20 traps to rodents and larger animals, however. Trap losses halted with the resumption Of ethylene glycol as a killing agent during the fOurth trapping period. Trap contents were collected at the end of each 48-hour trapping period by removing the inner cup and straining the contents through fine mesh. The contents were washed into a 120 ml collection jar using a solution of 95% ethyl alcohol and 5% glycerin. Specimen jars were labeled with trap number and collection date, then stored for subsequent sorting and identification. Beetle Sorting A morpho-classification technique was employed to create a type collection of pinned specimens based on Lindroth’s (1969) terminology. The 16 majority Of specimens were typed using this collection, then stored in sample vials with a solution of 95% ethyl alcohol and 1% glycerin. Samples were recorded by treatment, trap number, species, number Of individuals, andidate. The morpho—type collection was tentatively identified using voucher specimens in the Entomology Museum of Michigan State University. The resulting type collection was corrected by Dr. Robert Acciavetti of the USDA Forest Service, Morgantown, West Virginia. Voucher specimens will be lodged at the Carnegie Museum Of Natural History by Dr. Acciavetti. The corrected type collection was then used to reclassify morpho-typed samples. Species nomenclature follow the Bousquet and Larochelle (1993). Vegetation Sampling and Weather Understory vegetation was sampled every four weeks throughout the trapping seasons to characterize treatment differences. Herbaceous and woody ground flora (<1 m tall) species richness and relative ground coverage were sampled on treatment plots using 1 m2 quadrats located 2 m from trap centers on random azimuths in 1994 (different for each sampling). Species coverage was occularly estimated. During 1995, fixed-location, 1 m2 quadrats were established within 2 m of each trap center to assess frequency and cover Of ground flora. Density of understory woody species (>1 m tail) was sampled using 10 m2 subplots nested on the 1 m2 quadrats. This sampling protocol was designed to aIIOw vegetative comparisons among trap sites and among seasonal beetle-catch dates. The basal area of the overstory was determined using a 10- factor prism, with the trap cup as the plot center. Overstory data were expressed 17 as the basal area at each trap. Seedling and sapling data were expressed as the total number of trees per ha in two diameter classes ( s 1.9 cm dbh, 2.0-5.9 cm dbh). Weather data were obtained from the Kellogg Experimental Forest. Daily maximum and minimum temperatures, degree day (base 50°F) accumulation, and precipitation data were used to calculate corresponding sampling period variables. Average minimum and maximum temperatures were calculated for the 48-hour sampling period by averaging the corresponding temperatures of the three dates that defined the sampling period. For example, minimum daily temperatures for May 13, May 14, and May 15 were averaged to Obtain the average minimum temperature for the first sampling period in 1994. Cumulative precipitation was calculated from January 1 each year through trap closure on each sampling date. Incident precipitation was defined as total rainfall occurring during the two-week trapping period, from trap closure to trap closure. Degree day accretion was calculated as the total number of degree days occurring. during the two-week period beginning on the first date after a sampling period and ending on the day of trap Closure in the following sampling period; for example, May 16 through May 29. Cumulative degree days were calculated by totaling degree days accumulated from January 1 in each year. Humidity data were not available from the weather station. Statistical Analysis All statistical analysis used the SAS System computer statistical software, (Stokes, et al. 1995). A p-value of 0.05 was used to test treatment means for 18 significant differences. Vegetation data were analyzed using standard linear ANOVA techniques to test for differences among the three treatments, after first testing for normality and homogeneity of variance. Data for tree and understory sapling and seedling data were analyzed as a randomized complete block design using the SAS generalized linear model procedure. Comparison between years using ground flora data was not possible due to the different sampling procedures used. Ground flora data collected for 1994 were collected from plots randomly located at each sampling date; thus comparisons between dates and traps were not possible. These data were summarized as the year-long average percent coverage of woody and herbaceous ground flora in each treatment. The 1995 ground flora data were collected from fixed-location quadrats randomly located at the beginning of the sampling year to allow comparison between traps and treatments by date. Ground flora data were summarized separately as the total percent coverage in woody and herbaceous vegetation classes per quadrat for each sampling date for use in ANOVA split-plot design (plots split for dates) and in regression analysis. Measures of total annual percent coverage, absolute frequency, and relative frequency were calculated by summing across replications within each treatments. Relative frequency was calculated using the sum of absolute frequencies for all ground flora species. Average total annual percent coverage was calculated by dividing the total annual percent coverage for each species by the number of quadrats (n = 18). 19 Regression analysis was used with data for plot vegetation density and ground flora percent coverage to probe the relationship between vegetation characteristics and beetle activity. Vegetation data were incorporated as continuous variables in the logistic regression models for beetle activity data. In several cases, the resulting models stretched the acceptable limits for the proportional Odds assumption that is critical for categorical data analysis using Proc Logistic in SAS. Sparseness of the beetle activity data for some species prevented analysis with vegetation data for some dates. Overstory basal area and data for seedling and sapling density were assumed constant for all trapping dates for this analysis. Ground flora data from the six sampling dates in 1995 were matched with the corresponding beetle trapping dates. Observations for the intervening catch dates were derived by averaging the relevant variables for the two adjacent sampling dates. Thus, the analysis used ground flora values for eleven trapping dates. Regression models were fitted with and without variables for burn treatments, and with woody overstory and ground flora vegetation of variables separately, to test the relative contribution of vegetation data. Models incorporating ground flora data could only be analyzed for 1995 due to collection methods. Beetle catch data were analyzed using Categorical Data Analysis procedures for the SAS System (Stokes, et al. 1995). Data from pitfall trapping experiments has traditionally been analyzed using ANOVA procedures for linear, continuous, and additive variables, with the aid of an appropriate transformation 20 technique to stabilize the variance (Holliday 1991, Jennings, et. al 1986, Southwood 1978) and occasionally with the aid of data pooling. Data were not pooled and transformed to homogenize variance in this study tooavoid the problems of reduced power to detect significant differences, and the inability of traditional transformation techniques to normally distribute the data collected in this study. Fortunately, pitfall trapping data can also fit alternative distribution models and data classifications. Beetle catch data can be described as categorical, with observations for each trap falling into one or more categories or species. The distribution of such data may fit a binary distribution; observations frOm traps are either “successes” in the event of positive counts for a particular category, or “failures” when traps do not yield beetles. This characterization of trapping data allows easier analysis Of “zero-inflated” data sets by eliminating the need for logarithmic transformations. Relative abundance of beetles among the treatments was assessed by comparison of the total number of individuals trapped per treatment by date, as well as by species. Comparisons of the abundance of Carabid fauna as a whole between habitats are infrequent in pitfall trapping research, because data derived from pitfall trapping may reflect only the trap’s relative efficiency in trapping different species (or the resistance of the ground vegetation to each species movement), rather than absolute population densities (Greenslade 1964). Yet, total Carabid relative abundance may be relevant when used as an index for comparing Carabid activity within the same habitat, across similar . 21 habitats, or between years (Mitchell 1963, Niemela et al. 1990. In spite of the differences in ground cover among the treatments, total Carabid activity should serve as an index of the change in beetle activity as a result of prescribed burning. Only species present in sufficient quantity for these analysis (greater than three percent of total annual catch) were analyzed, as determined by trial and error. Separate analyses were conducted for each year, out Of concerns that trapping data may reflect the random effects of two different causal events on two-year-burn treatments; the fires in 1993 and in 1995. Variance was analyzed using logistic regression analysis, with beetle count data treated as the categorical response variables. All but one species specific analysis used only two classes for the response variable; zero beetles, and greater than zero beetles. The analysis for Pterostichus stygicus (Casey) used three count classes. Total beetle activity analysis used six classes to characterize count data. Table 1 describes the count data classes used for each Of the seven Carabid species. Because of the sparseness of beetle data on some dates, only sampling dates with sufficient observations (2 or more beetles trapped per sample date) to allow analysis were used in regression analysis. Table 2 summarizes the sampling dates used for each species analyzed. Variance in beetle abundance attributable to time, topography, and other edaphic features were apportioned by including the terms for date, blocking, and seasons in each model. Total relative abundance of Carabids was analyzed using data for all trapping dates, and terms for blocking and seasons. Blocking for seasons was arbitrarily designated after examining plots of trapping date vs. v— ' " ' 'u -_'.L__ 22 Table 1. Number of beetles in count data classes used to categorize the relative activity of seven Carabid beetle species for logistic regression analysis. Data Class Species 1 2 3 ‘ 4 5 6 Cyclotrachelus convivus (Leconte) 0 21 C. soldalis (Leconte) 0 21 Carabus goryi Dejean 0 21 Pterostichus mutus (Say) 0 21 P. stygicus (Say) 0 1 >1 P. tn‘stis (Dejean) 0 21 Tn‘chotichus vulpeculus (Say) 0 21 Total Carabid Species 0 1 2 3-4 5-8 9-15 Table 2. Sampling dates"2 used for logistic regression analysis for each of the seven most active Carabid species. Species Beetle Sampling Date Cyclotrachelus convivus (Leconte) 4 through 8 C. soldalis (Leconte) 4 through 8 Carabus goryi Dejean 1 through 8 Pterostichus mutus (Say) 1 through 3 P. stygicus (Say) 4 through 12 P. tn’stis (Dejean) 4 through Tn'chotichus vulpeculus (Say) 4 through Total Carabid Species 1 through 12 1Sparseness of count data for some beetle species did not allow use of all sampling dates with logistic regression analysis procedures. 2Dates chosen for analysis correspond to the rough season classification outlined in Figure 1, and contain sufficient observations to allow analysis. 23 degree day accretion. Seasons were assigned to three periods with roughly equal degree day incidence rate, separated at inflection points indicated on Figure 3. Individual species activity varied according to date; some species were not present in trap counts until mid-summer, others were present only in spring. Thus, species specific activity data were analyzed using blocking variables for topography, and the dates present. Species Diversity Analysis Common diversity indices measuring species richness, abundance, and evenness were calculated, treating each treatment-year combination as a distinct community. This procedure followed suggestions by Magurran (1988) and Southwood (1978) for comparisons of diversity among habitats. Species diversity measures were calculated using the total number of beetles per treatment in each species. Shannon, Simpson, Maraglef, and Berger-Parker diversity indices were calculated for each treatment. The Shannon index was selected because it is widely used and may provide a basis for comparison with other studies. It also has moderate discriminate ability and sensitivity to sample size. Differences in the Shannon H’ indices were tested using t-tests as described by Hutcheson (1970). The Maraglef index reflects species richness and has good discriminate ability, but it is highly sensitive to bias induced by small sample size. The Berger-Parker index reflects species dominance and has low sensitivity to sample size. The Simpson index also measures species dominance, but it has better discriminate ability. The Shannon evenness index 24 SAIIelnIunQ eaten. giant. 2:335 NF 3 or m w R m m v m N F o 2 2 2 _ u a . o _ 9mm 98me m>:m_:E:o vmmvi - on com I 33 0230 02:23:50 mam? 0' \D e I / 22328.4. 25 $58 33. .ol . cow A I cozmbo< >8 moaoo vmmviqi ooor . - one oomw I \ - CON \ 88 - \\ - 8N . \ . oom comm . - omm | L coon . \ I 8... /c \ macaw __mu_ LOEEsm comm omv .33 E 2 30-3 >22 use 39 :_ we .30-: >22 excess votsooo can 930: 3. 2o; meeton 9.3%.; .E om omwnv mmmr can 33 5 5:53:50: uca 5:203 >3 35% he not: szfifi co comes wcommom BE moored 9:363 ho cozaxLano .n 2:9". uonamv 25 was also calculated; it has poor discriminate ability, but is only moderately sensitive to sample size. The log series a from Williams’s nomograph (Williams, 1964; Southwood, 1978) also was calculated. Species abundance models have been fitted for the geometric series, the log series, the log normal series, and the broken stick models. Goodness-of—fit for each abundance model was calculated using the X2 test and a p-value of 0.05. Chapter 3 RESULTS AND DISCUSSION Overstory and Woody Understory The mixed pine species overstory in the plantation appears to be relatively homogenous with respect to species density and composition. Average red and white pine densities were 131 and 93 trees per ha, respectively. Average DBH’s ranged from 16 to 40 cm for red pine and from 20 to 60 cm for white pine, with the largest trees of both species found on the lower slope positions. The three treatments also contained similar average basal areas of red and white pine, about 29 and 16 m2 per ha respectively (Table 3). Results of the one-way analysis of variance for overstory basal area indicate no significant differences in density of the overstory among treatments (Pr F > = 0.81) and blocks (Pr > x2 = 0.74) despite a history of variable thinning. The effects of periodic burning were readily apparent in treated areas within the ground flora, seedling, and sapling vegetation classes. Vegetation on the 2-year burn plots was composed chiefly of low herbaceous and woody ground flora. Only three saplings (2.0-5.9 cm dbh) were observed in eighteen 10 m2 sample plots; two observations occurred near the edges of treatment 26 27 Table 3. Mean overstory basal area (mzlha) among burning treatments. Treatment means are followed by their standard deviation. Burn Interval Mean WP1 RP1 2-year 47.5 i 11.2 17.8 29.1 6-year 46.7 i 8.4 15.2 30.1 Control-—no burns 48.8 i 7.9 16.9 29.6 1WP = white pine, RP = red pine 28 areas, and one sapling occurred in an area of the plot where fire did not carry well. These unusual observations were considered outliers and excluded from further analysis. These plots also were virtually devoid of seedling-size woody vegetation (0-1.9 cm at dbh). Nevertheless, ANOVA’s for sapling and seedling data were included all burn treatments. Saplings and seedlings were most dense on the unburned plots. Control treatments were well stocked with advanced regeneration, averaging 3,944 stems per ha in the 2.0 to 5.9 cm dbh class. Saplings were scarce in the once- burned, 6-year treatment plots, with fewer than 400 stems per ha (Table 4). Analysis of variance indicated significant differences in sapling density among treatments (p = 0.0001) but not among blocks (p = 0.34). Multiple comparison tests indicated that differences were significant at p= 0.05 between burned and unburned treatments, but not between burn treatments. Seedling density on the 6-year-burn plots was roughly half the density of the unburned stand areas (Table 4), with roughly 9,300 and 16,100 seedlings per ha, respectively, while no seedlings were observed during sampling of 2- year burn plots. ANOVA indicated significant differences in seedling density among the treatments (p = 0.0001 ), but not among blocks (p = 0.08). Multiple comparison procedures indicated significant differences among all three treatments at p = 0.05. These data are similar to the results of prescribed burn studies in Northern Lower Michigan red pine stands. Henning and Dickmann (1996) 29 Table 4. Seedling (_<_ 1.9 cm at dbh) and sapling (2.0-5.9 cm dbh) density per ha in burned and unburned treatments. Means for each variable are followed by their standard deviation. Burn Interval Seedlings Saplings 2-year 0.0 i 0.00 1661 i 1,130 6-year 9,277 i 7,870 277 i 960 Control-no burns 16,111 1 9,010 3,944 i 1,860 1These estimates reflect data outliers; saplings growing at the edges of plots or in incompletely burned areas. 30 reported sapling densities of 210, 385, and 4,336 stems per ha for 2-year burns, 5-year burns, and unburned treatments, respectively. They also noted significant differences among these treatments for seedling density, following the same pattern. Niering (1970) and Hodgkins (1958)reported that frequent prescribed burns kill most stems < 10 cm dbh, while larger trees escape relatively unscathed. Prescribed fire effects are short-lived in the absence of repeated burning, as hardwood regrowth may reach heights of up to 1.8 m after only 3 growing seasons (Hodgkins 1958). The qualitative effects of burning were most evident in the seedling-size class (<1.9 cm dbh). Six-year burn plots contained greater density and a higher percentage composition of disturbance-favored species than the control treatment, including common buckthorn (Rhamnus cathartica), sassafras (Sassafras albidum), Viburnum spp., and tulip poplar (Un'odendron tulipifera) (Tables 5, 6, and 7). Control plots contained greater percentages of thin barked species and shrubs, including black cherry (Prunus serotina), red maple (Acer rubrum), and bush honeysuckle (Diervilla Ionicera). Several species occurred only on unburned plots: hackberry (Celtis occidentalis), flowering dogwood (Cornus florida), musclewood (Carpinus caro/iniana), and prickly ash (Zanthoxy/em americanum) (Table 6). These plots were dominated by black cherry, honeysuckle, red maple and sugar maple (Acer saccharum) seedlings. Prescribed burning apparently did not result in increased oak seedling establishment; control plots contained more red oak (Quercus rubra), white oak (Quercus alba), and black oak (Quercus velutina) than 6-year burn plots. ._. .. 31 Table 5. Seedling species composition of 6-year burn and unburned control plots as the percent of total seedling density within each treatment ( 31.9 cm dbh size class).1 Bum BC2 BU EBU ELD HON RM RO SAS SM VIB Other Total Interval spp. 6-year 5% 19% 5% 5% 0% 9% 7% 26% 7% 5% 12% 100% Control- 20% 6% 5% 3% 8% 22% 4% 3% 7% 0% 22%100% no burns TSpecies present at 25% in any treatment. No seedlings were recorded in plots burned at 2-year intervals. 2Species abbreviations: BC-black cherry, BU-common buckthorn, EBU-European buckthorn, ELD-elderberry, HON-bush honeysuckle, RM-red maple, RO-red oak, SAS- sassafras, SM-sugar maple, VlB-vibumum 32 Table 6. Mean seedling ( 31.9 cm dbh) species density (per ha) in burned and unburned plots. Burn Interval1 Species 6-year Control- no burns Acer negundo 56 56 Acer rubrum 833 3,61 1 Acer saccharum 667 1 ,1 1 1 Carpinus caro/iniana 0 56 Carya spp. 167 722 Celtis occidenta/is 0 111 Cornus flon'da 0 56 Cory/us cornuta 111 167 Diervilla lonicera 0 1,278 Fraxinus americana 111 333 Fraxinus pennsy/vanica 0 333 Liriodendron tulipifera 222 0 Morus spp. 56 111 Prunus avium 56 167 Prunus serotina 500 3,167 Quercus a/ba 389 667 Quercus rubra 667 667 Quercus velutina 0 333 Rhamnus cathartica 1,722 944 Rhamnus frangu/a 444 778 Sambucus spp. 444 444 Sassafras albidum 2,389 556 Viburnum spp. 444 0 Vitus spp. O 389 Zanthoxy/um americanum 0 56 1No seedlings were observed in the 2-year burn interval plots. 33 Table 7. Species listed in “Other” category (present at <5% of treatment density) in Tables 5 and 8. Seedlings Saplings Acer negundo Cornus florida Carpinus caroliniana Diervilla Ionicera Carya spp. Fraxinus pennsylvanica Celtis occidentalis Popu/us tremuloides Cornus florida Quercus rubra Cory/us cornuta Rhamnus cathartica Fraxinus americana Rhamnus frangu/a Fraxinus pennsylvanica Liriodendron tulipifera Morus spp. Prunus avium Quercus velutina Vitus spp. Zanthoxylum americanum 34 Sapling species composition in the 6-year-burn and control plots also appears to reflects the frequency of disturbance. Six-year—burn plots contained only rapidly growing, intermediate or intolerant shade Species including sassafras, hickory (Carya spp.), red maple, and black cherry (Tables 8 and 9). Burn plots notably lacked the more tolerant species found in control plots; sugar maple, green ash (Fraxinus pennsylvanica), flowering dogwood, and white ash (Fraxinus americana). Sassafras appears to be the most important fire-favored hardwood species on this site. Observations on 2-year burn plots prior to re-buming in May of 1995 suggested that sassafras had very high density in the most frequently burned plots. Although few saplings survived the 1995 spring burn on 2-year plots, root suckers of sassafras quickly re-sprouted and were nearly ubiquitous after only 2 months, although they didn’t grow tall enough to be included in the woody understory seedling class. The 6—year burn plots contained nearly 4 times the “seedling” stems per ha as unburned plots. This species has similar responses to those noted for black cherry in a northern Michigan study. Henning and Dickmann (1996) noted that black cherry density was directly related to frequency of burning in red pine stands, possibly owing its success on frequently burned plots to a seed bank strategy. Black cherry has hard coated seeds that are readily disseminated by birds. These seeds undoubtedly survive in the forest floor, germinating quickly following fire. The same response from black cherry was not apparent in my study, probably due to 35 Table 8. Sapling species composition of 6-year burn and unburned control plots as the percent of total seedling density within each treatment (2-5.9 cm size class).1 Burn Interval BC2 HIC RM SAS SM WA Other Total spp. 2-year3 33% 0% 0% 0% 67% 0% 0% 100% 6-year 20% 20% 20% 40% 0% 0% 0% 100% Control-no burns 21% 3% 45% 7% 4% 6% 14% 100% 1Species present at 25% in any treatment. 2Species abbreviations: BC-black cherry, HIC-hickory, RM-red maple, SAS-sassafras, SM-sugar maple, WA-white ash 3These estimates are based on three saplings tallied from (n=18) 10 m2 plots. These are considered data outliers and have been discarded for ANOVA’s and logistic regression analyses. 36 Table 9. Mean sapling (2.0-5.9 cm dbh) species density (per ha) in burned and unburned plots. Burn Interval Species 6-year Control- no burns Acer rubrum 56 1 ,778 Acer saccharum O 167 Carya spp. 56 111 Cornus florida 0 56 Diervilla Ionicera 0 56 Fraxinus americana 0 222 Fraxinus pennsylvanica 0 56 Popu/us spp. 0 56 Prunus serotina 56 833 Quercus rubra 0 56 Rhamnus cathartica O 222 Rhamnus frangu/a 0 56 Sassafras albidum 111 278 37 differences in site quality and the presence of more rapidly growing competitors at the Kellogg Forest. Ground Flora Burned plots contained a rich community of groundflora species, immediately discernible from unburned stand areas. Control plots were relatively depauperate in the diversity and development of understory flora; these areas were characterized by sparse low herbaceous and woody seedling growth. Two-year burn plots contained thick undergrowth that generally averaged less than 1 m in height. Six-year burn plots contained a greater composition of woody ground flora species in 1995, but appeared similar to 2- year plots in 1994; ground flora here averaged 1 to 1.5 m in height. Coverage by woody and herbaceous ground flora components in 1 m2 frames was consistently higher on burn treatment plots than on unburned plots in 1994 and 1995 (Tables 10—13). Average percent coverage estimates appear similar for 2-year-burn and 6-year-burn plots in both years. Remarkably, compared to 1994, 1995 coverage and relative frequency of woody ground flora vegetation only decreased slightly (average of 5 percent) on 2-year-burn plots following re-burning of that treatment in May. Total average coverage also appears similar between years; unburned plots had 27 and 30 percent coverage in 1994 and 1995, respectively. Total coverage for the 6-year burn was nearly the same over the two years. Compared to 1994, total coverage for 1995 two-year burn plots decreased only slightly from 57 to 52 percent, despite extensive consumption of surface litter 38 Table 10. 1994 woody ground flora species mean percent coverage by burn interval. Woody Species 2-year 6-year Control- no burns Acer spp 0.4 1.6 2.1 Carya spp. 0.4 0.5 0.3 Diervilla Ionicera 0.4 0.4 0.6 Fraxinus spp. 0.2 0.3 0.2 Parthenocissus quinquefo/ia 1 1.1 12.0 4.9 Pinus spp. 0.3 0.2 0.7 Prunus spp. 0.3 0.4 1.1 Quercus spp. 0.9 1.9 0.5 Rhamnus cathartica 1.5 1.9 0.5 Rhamnus frangula 0.3 0.0 0.0 Rhus spp. 0.2 0.3 0.0 Ribes cynosbati 0.0 0.0 0.1 Rosa spp. 0.0 0.1 0.0 Rubus spp. 16.7 12.7 2.7 Sambucus spp. 0.5 0.0 0.1 Sassafras albidum 4.8 4.8 2.7 Smilax hispida 0.9 0.2 0.9 Tilia americana 0.0 0.1 0.5 Toxicodendron radicans 1.3 1.4 0.7 Vitus spp. 0.4 1.8 1.4 Woody Species Total 40.7 40.6 20.0 s.d. 1:37.97 i33.49 124.11 39 Table 11. 1994 herbaceous ground flora species mean percent coverage by burn interval. Herbaceous Species 2-year 6-year Control- no burns Apios amen’cana 1.1 1.6 0.3 Apocynum androsaemifolium 0.1 0.8 0.0 Aralia nudicaulis 0.0 0.0 0.0 Arctium minus 0.1 1.0 0.0 Aster spp. 0.1 0.9 0.1 Carex spp. 0.1 0.3 0.0 Circaea quadn'sulcata 1.1 0.3 0.6 Cirsium spp. 0.1 0.1 0.0 Fragan’a virginiana 0.1 0.2 0.0 Desmodium glutinosum 0.4 0.2 0.0 Floerkea proserpinacoides 0.0 0.0 0.0 Gallium spp. 0.3 0.5 0.4 Geranium maculatum 0.0 0.1 0.0 Hieracium aurantiacum 0.1 0.3 0.0 Hypericum punctatum 0.0 0.0 0.0 Lactuca scan'ola 0.1 0.1 0.0 Maianthemum canadense 0.0 0.1 0.0 Mitchel/a repens 0.0 0.3 0.0 Moss 0.0 0.0 0.0 Osmarhiza claytoni 0.1 0.3 0.2 Oxalis acetosel/a 0.2 0.1 0.0 Phrymu Ieptostachya 0.2 0.3 0.2 Phytolacca americana 4.8 1.3 0.1 Pilea pumila 0.4 0.0 0.3 Plantago major 0.0 0.0 0.0 Poa spp. 0.2 0.4 0.0 Podophyllum peltatum 0.0 0.2 0.0 Polygonatum biflorum 0.3 0.6 0.3 Potentilla simplex 0.1 0.0 0.0 Pten‘dium aqui/inum 3.2 4.4 2.9 Rumex acetosella 0.1 0.1 0.0 Taraxacum officinale 0.1 0.1 0.0 Trillium spp. 0.1 0.0 0.0 Unknown #10 0.1 0.0 0.0 Unknown #11 0.0 0.0 0.0 Verbascum thapsus 0.0 0.0 0.0 Viola spp. 2.7 0.7 1.7 Herbaceous Species Total 16.4 15.4 7.4 s.d. 1:22.44 1:20.03 1.42.44 40 Table 12. 1995 woody ground flora species mean percent coverage by burn interval. Woody species 2-year 6-year Control- no burns Acer spp. 0.2 1.5 2.4 Berberis thundbergii 0.0 0.3 0.0 Carya spp. 0.0 0.7 0.0 Celastrus scandens 0.5 0.0 1.8 Diervilla Ionicera 0.0 0.6 0.5 Fraxinus spp. 0.0 0.0 0.0 Parthenocissus quinquefolia 0.0 0.0 3.4 Pinus spp. 6.6 13.1 0.7 Prunus spp. 0.1 0.1 1.4 Quercus spp. 0.0 0.8 0.0 Rhamnus cathartica 0.2 1.0 0.1 Rhamnus frangu/a 0.1 0.8 0.1 Rhus spp. 0.0 0.0 0.0 Ribes cynosbati 0.0 0.0 0.2 Rosa spp. 0.0 0.0 0.0 Rubus spp. 15.6 13.3 4.2 Sambucus spp. 0.9 0.2 0.6 Sassafras albidium 4.7 3.8 0.5 Smilax hispida 0.3 0.0 1.4 Solanum dulcamara 0.0 0.2 0.0 Tilia americana 0.0 0.0 0.0 Toxicodendron radicans 2.5 1.8 1.1 Vitus spp. 0.5 1.6 2.2 Woody Species Total 32.1 39.9 20.7 s.d. £31.94 1:36.52 i25.58 41 Table 13. 1995 herbaceous ground flora mean percent coverage by burn interval. Herbaceous Species 2-year 6-year Control- no burns Apios americana 0.3 0.9 0.6 Apocynum androsaemifolium 1.0 0.0 0.0 Aralia nudicallis 0.0 1.6 0.0 Arctium minus 0.0 0.2 0.0 Aster spp. 0.1 0.2 0.2 Carex spp. 0.0 0.0 0.0 Circaea quadn'sulcata 1.3 0.2 0.8 Cirsium spp. 0.1 0.0 0.0 Fragaria virginiana 0.0 0.3 0.0 Desmodium glutinosum 0.6 0.0 0.0 Floerkea prose/pinacoides 0.0 0.0 0.0 Gallium spp. 0.4 0.5 0.7 Geranium maculatum 0.0 0.0 0.0 Hieracium aurantiacum 0.0 0.5 0.0 Hypen'cum punctatum 0.0 0.0 0.0 Impatiens biflorum 0.0 0.0 0.0 Lactuca scario/a 0.1 0.0 0.0 Maianthemum canadense 0.0 0.2 0.1 Mitchel/a repens 0.0 0.0 0.0 Moss 0.0 0.0 0.2 Osmarhiza Claytoni 0.0 0.5 0.3 Oxalis acetose/la 0.1 0.1 0.0 Phrymu Ieptostachya 0.1 0.2 0.3 Phytolacca americana 7.8 1.8 0.0 Pilea pumila 5.0 0.0 0.2 Plantago major 0.0 0.1 0.0 Poa spp. 0.1 0.2 0.0 Podophyllum peltatum 0.0 0.0 0.0 Polygonatum bif/orum 0.1 1.0 0.4 Potentilla simplex 0.0 0.0 0.0 Pten'dium aquilinum 2.0 7.6 3.0 Rumex acetose/Ia 0.0 0.1 0.0 Taraxacum officina/e 0.1 0.1 0.0 Trillium spp. 0.0 0.0 0.0 Unknown #10 0.0 0.0 0.0 Verbascum thapsus 0.0 0.0 0.0 Viola spp. 0.3 1.0 2.6 Herbaceous Species Total 19.6 17.5 9.6 s.d. i268 i241 i101 42 and scorching of woody stems by surface fire during re-burning of the plots in May. Differences in 1995 total coverage means for herbaceous and woody vegetation components were significantly different for treatments (F = 7.46, p = 0.0007) and blocking (F = 11.81, p = 0.0001). Herbaceous ground flora means for 1994 and 1995 were not significantly different between 2-year and 6-year burn treatments in Student-Newman-Keuls and Tukey's HSD ranked means tests (alpha = .05) (Tables 14 and 15). Differences between 2-year and 6-year burn treatment woody ground flora coverage means were not significant in 1994, but significant differences between 2-year burn and 6-year burn means were found using Student-Newman-Keuls tests, but not Tukey’s HSD (Table 15.). The Student-Newman-Kuels test has fairly good power to detect differences between means, but does not control the maximum experiment-wise error rate (Einot and Gabriel 1975). Tukey’s HSD controls the type | experiment—wise error rate, but has weaker power to detect type II errors (Hayter 1984). Differences in ground flora species richness were clear in both years between unburned control treatments and 6-year burn plots in both years, but not so with 2-year burn treatments (Table 16). Control treatments contained fewer species in 1994 and 1995 than 6-year burn plots. Two-year burn plots had greater species richness than unburned plots in 1994, but four fewer species in 1995. Following spring burning in 1995, 2-year plots had 11 fewer herbaceous species than in 1994. Given that the 1994 2-year plot vegetation data represented the second year since burning, it is logical to assume that the Table 14. Ranked means tests for 1994 herbaceous and woody ground flora percent coverage. 43 Burn Interval Herbaceous 2-year 6-year Control- no burns Mean 16.4 15.4 7.4 SNK1 a a b Tukey’s HSD‘ a a b Woody 6-year 2-year Control- no burns Mean 40.7 40.6 20.0 SNK2 a a b Tukey’s HSD2 a a b 1means with the same letter are not significantly different (alpha = 0.05, df = 249, n = 90). 2means with the same letter are not significantly different (alpha = 0.05, df = 249, n = 90). 44 Table 15. Ranked means tests for 1994 herbaceous and woody ground flora percent coverage. Burn Interval Herbaceous 2-year 6-year Control- no burns Mean 19.6 17.5 9.6 SNK1 a a b Tukey’s HSD1 a a b Woody 6-year 2-year Control- no burns Mean 39.9 32.1 20.7 SNK2 a b c Tukey’s HSD2 a a b 1means with the same letter are not significantly different (alpha = 0.05, df = 300, n = 108). 2means with the same letter are not significantly different (alpha = 0.05, df = 270, n = 108). Table 16. Number of ground flora species by year and burn interval. 1994 1995 Vegetation 2-year 6-year Control- 2-year 6-year Control- Type no burns no burns Herbaceous 30 32 22 19 26 19 Woody 18 17 18 12 18 16 Totals 48 49 40 31 44 35 45 1995 fire temporarily reduced species richness on 2-year burn plots; the herbaceous fauna should recover to near 1994 richness by the end of 1996. Studies by some researchers suggest that ground flora communities exhibit distinct differences in density and species richness following repeated prescribed burning. Henning and Dickmann (1996) found that ground flora coverage on biennially burned treatments in a northern Michigan red pine stand was lower than treatments with longer intervals between fire; plots with 5 and 10 year burn frequencies had the greatest herbaceous coverage. However, species richness did not differ among burn treatments and between burned and unburned plots. By contrast, White (1983) noted that repeated prescribed fires increased overall species richness in a Quercus ellipsoidalis community in Minnesota. Lemon (1949) showed that ashes on burned sites stimulate a lush early herb and shrub growth, although survival and increase of herbs are related to life history and form. Woody species richness differed little among treatments and years (Table 16). Woody species on 2-year plots in 1995 suffered topkill during burning, but the majority of species quickly re-sprouted or germinated from buried seed. The lower overall species richness observed in 1995 was probably due to the difference in plot locations during successive years. The woody ground flora of control plots contained greater representation of species easily killed by fire or with higher shade tolerance than found on burn plots (Table 12). Control plots had the greatest average percent coverage by maple seedlings, bush honeysuckle, pine species, American basswood (Tilia 46 americana), vine bittersweet (Celastrus scandens), wild current (Ribes spp.), greenbriar (Smilax hispida), and in 1995, grapevine (Vitus spp.). Two-year burn plots were characterized by fast-growing, disturbance favored species. Although most of the woody stems were killed by the surface fire during 1995 spring burning, these plots were quickly re-vegetated by a dense growth of blackberry (Rubus spp.), sassafras, and poison ivy (Toxicodendron radicans). By mid-summer, blackberry and sassafras on the freshly burned plots had grown to about 1 m in most areas. Biennially burned plots had greater average coverage by blackberry, sassafras, and elderberry (Sambucus spp.) in both years, and the least average coverage in species noted on control plots. Six-year burn treatments contained the greater average coverage by species that re-sprout after top-kill by fire, i.e. oak and hickory, or that are ~ disturbance-favored but relatively slow growing, i.e. Virginia creeper and barberry (Berben‘s thundbergii). These plots also contained a significant representation by species whose fruit are spread readily by birds and rodents, like common buckthorn, sassafras, and in 1995, black cherry. Burned treatments contained greater overall coverage by almost all herbaceous species encountered than unburned plots, although mean differences were not statistically tested at the species level. Control treatments contained more moss, bedstraw (Gallium spp.), and in 1995, violets (Viola scandens), than burn treatments. Herbaceous flora in burned treatments reflect the frequency of disturbance, as well as the degree of shading by woody plant 47 species. Two-year burn plots in 1995 (season immediately after spring burning) were dominated by pokeweed (Phytolacca americana), nettle (Pi/ea pumila), and enchanter’s nightshade (Circaea quadrisulcata). Species coverage on these plots in the second season after prescribed burning (1994) appeared more like that of the 6-year burn plots, with greatest coverage by pokeweed, bracken fern (Pteridium aquilinum), tretick foil (Desmodium glutinosum), and violets. The 6- year burn plots contained greater coverage of the following species in both years: fern, common burdock (Arcticum minus), wild sarsaparilla (Aralia nudicallis), Solomon seal (Polygonatum biflorum), groundnut (Apios americana), sweet cisely (Osmarhiza claytonr), mayapples (Podophyllum peltatum), and Aster species. Weather Effects Weather conditions during the 1995 were hotter and drier than observed in 1994. Data collected from Kellogg Experimental Forest’s weather station also indicate that the winter of 1994 was much colder than 1995. Average minimum temperatures in January and February of 1994 were less than -14 C, with minimum temperatures well below -17 C for more than a week. Temperatures were relatively balmy in early 1995, with minimum readings above -17 C and averages above -7 C. This trend continued throughout the year (Figure 4); average temperatures were 4.5 C (10 F) higher during the 1995 trapping season than in 1994. Minimum temperatures averaged 3 C (8 F) higher during 1995. Degree day accumulation in 1995 was 150 degree days (base 50 F) ahead of 1994 levels at the start of pitfall trapping in May (Figure 5). Cumulative 48 degree days averaged 680 days ahead of 1994 throughout the season, reaching 875 degree days by the end of trapping in 1995. The greatest difference in degree day incidence occurred in late June through August sampling dates. Incidence in fall dropped to near or below 1994 levels, when temperatures for the two years were similar. Overall, 1995 was a slightly drier year than 1994, with the largest differences occurring in snowfall and mid-season rainfall. Pre-trapping season rainfall was similar, with 24.3 and 21.1 cm in 1994 and 1995, respectively. Nearly twice as much snow fell in 1994; 80.6 cm compared to 48.9 cm in 1995. Cumulative precipitation in 1995 lagged 2 to 8 cm behind 1994 levels throughout the trapping season, averaging 6.3 cm less than in 1994 during summer and fall (Figure 6). Rainfall occurred more frequently in 1995 than in 1994. Unusually heavy rainfall occurred during June and August of 1994, with low levels of precipitation during the rest of the year. 49 oozed ascent. 33.0255 NF 3 or a m N. o n v m N r — l_, _ _ V _ h 2 F V _ _ _ _ a a . 2 1.1 . meaflanok SEES—2 mmmFIII meafloquk E2555. 3.3 I 8388:.8. 09:03. m8 Flt 2388th 09803. 33 u e. . on .2320 out 3 mEcoao So: fierce giant 2.: ante—coo «“5 £36 095 05 .6.— mEEEoQEE mom—20>“ 05 3:39.62 «Eon seam .33 a 33 co.— mcomuom 9.3%.: acts“. 3.32888 EEEEE uca emaco>< .v 959”. (o) ammedwal 50 awgmnmno vote... 559...... 23.335 we _. F o r o o x. o w v m N F o . u .r . .. . . u 1 _. _ o .960 $.95 6263.56 3310'. \ $60 @0300 0253830 moor II .T on com 1 | ‘1. 5.6.0.4 >8 $.95 moor . 0| / 5.6.694 .80 06.95 33+ \1 o3 I ooow : \ \ .- o2 82 -. \o , a \ . oom 1 omN Bow .1 a \ \ s /p \ s / \ -- com 88 + \u \ 2 4 \VA ( 1 8m \ \ I I ‘0‘ 88 ul- \- /- \ -- 8.. oomm omv .E on 82.. 39 can 33 :. common 9.3%.. 05 9:56 cozsaanoa can 630.com >2. 00.03 .m 0.50.“. uogzamv 51 (um) uonendgoaId IuapgouI 02.00 9.330; 2.0035 N. S S 0 0 .. 0 0 4 n N F o. . m I I1 L. n . . . n I I I o N n I i I - I s a .. N .t I ‘ \ f I ‘ OF — \ V ( a — — s - V 1r ’ 1r ON — s - ‘ d i 8 h o 1 . . a \ -. on a u a C h w i a h \ u 1 av .3265 32+ . . a\ . h 2 -- $022.32 I I . \I. I... a . .- 8 03.03530 32 II . 4 . . 02.0358 03.101 . 4 \\ a . N? -1 a h t 1 ow - .. \ . u 3 -. \II III I\ c -- E ‘ \ 2 8 .0300 9:330: 9.3.0:. 0:0 05000003 03.0.. x00? 93 05 55.30 @5530 3.003.003 .o 0m0.0>0 05 0300030. 3.03 :00m .003 0:0 33 :. 0:00000 9.330.. 05 .06 3:83.003 0:00.05 x003-~ 0:0 02.0.:an .o 0.39.“. (tun) uoueudgaaId eAgmnwng ”"7 “'1‘?— 52 Carabid Beetle Relative Activity Pitfall trapping catches indicate that fire and fire frequency have a pronounced effect on beetle activity. Beetles were more active on frequently burned than unburned treatments in both years (Figures 7 and 8). Nearly twice as many Carabids were caught on 2-year burn plots than on unburned plots, 443 beetles in 1994 and 378 in 1995, compared to 233 and 177 for unburned plots, respectively. Traps on six-year burn plots caught more beetles than control plots in 1994, but not in 1995 ( Table 17). Total catches over two years clearly show that fires at short intervals had a major effect on beetle activity, whereas fires at 6 year intervals had little effect. The effect of burn treatment is clearly demonstrated at the species level also (Tables 17 and 18), although the trend varies for each species and for each year. The most commonly trapped species, Pterostichus stygicus (Say), was least abundant in traps on the 6—year burn plots in both years. Synchus impunctatus (Say) was trapped more often in 6-year burn plots in 1994 than in other treatments. Activity patterns of four species changed in 1995; all showed decreased activity on 6-year burn plots, three showed higher activity on control treatments than 6-year burn plots that year. Three species exhibited activity that seemed directly related to frequency of burning in both years: Pterostichus mutus (Say), Carabus goryi Dejean, and Cyclotrachelus soldalis (Leconte). Similar results have been documented by Harris and Whitcomb (1971), who reported higher abundance among four species of Pterostichus (formerly Evarthrus: Carabidae) in burned compared to unburned southern pine stands. 53 N. 02.03 9.330.... 20.00.55 0E3 o:-_o..:oo I I .00»-.. 0| .mm>-NlIl .32. E 02.03 9.330.. 3000 9:50 00330.. 02.003 .3 .0353: 3.0... N 0.39.... S. sanaaq lo JaqumN rr-._ .2 54 02.03 3:330; 20.0035 : or m w h o m 0 m N _. _ w w .r 0 L n . 1. w w o \l . / l/ u I I\... VII I I L ( III II. \ ‘ \ \ ) 11 Or a \ N ./ \ . . « \x s a /« s /o\ -- cm 1 a 0 . . / fl § I: On 0 at I I t 00 1 8 0353 2.32.80 I I .00>-o CI -1 co .00>-NIIIII on .003 E 02.03 9.330.. :000 9...:0 00330.. 02.003 .0 .0353: .0.o.P .0 0.093 sanaaq Io JaqwnN 55 Table 17. Total number of Carabid beetles trapped by species and burn treatment. Control 6 year burn m Species 1994 1995 1994 1995 1994 1995 Sphaeroderus stenostomus Iecontei Dejean 1 9 1 0 1 5 4 10 4 Carabus goryi Dejean 5 22 21 32 35 Notophilious spp. 0 0 0 2 Pterostichus femoralis 0 0 0 2 P. melanarius (llliger) 1 6 O 8 P. mutus (Say) 6 7 62 39 P. pensylvanicus LeConte 1O 0 12 4 P. pennundus (Say) 2 3 2 3 P. stygicus (Say) 110 6 96 3 137 88 P. tn‘stis (Dejea n) 16 1 14 1 8 14 Poecilus lucublandis (Say) 1 8 2 4 Cyclotrachelus convivus (LeConte) 13 1 14 34 26 C. soldalis (LeConte) 5 7 25 30 Calathus gregarius (Say) 0 3 2 3 Synuchus impunctatus (Say) 23 63 53 1O Cymindis americanus Dejean 2 2 Agonum ferreum Haldeman 0 0 A. melanarium Dejean Platynus h ypolithos (Say) 1 Galerita janus (Fabricius) Pseudoamara arenaria (LeConte) Amara familian’s Duftschmid A. impuncticollis (Say) Amphasia interstitialis (Say) Harpalus compar LeConte H. erythropus Dejean H. fulvilabris H. herbivagus Say H. Iongicollis LeConte H. paratus LeConte H. pensylvanicus (DeGeer) H. providens Casey H. puncticepts (Stephens) H. somulentus Dejean Anisodactylus nigerriumus (Dejean) A. rusticus (Say) Notiobia nitidipennis (LeConte) Tn'chotichus dichrous (Dejea n) 7'. vulpeculus (Say) Bradycellus badipennis (Haldeman) Chlaenius tricolor Dejean Dicaelus elongatus Dejean Unknown 1 Unknown 2 Unknown 3 teneral spp. Total per year 233 177 295 140 444 37 Total for 2 years 410 435 882 N m-5CD-*f\.)(DCDC>(O-3OOOC>-¥\lo--'~—¥C3l\JCDCDOO-lO-‘-hl\J(.I~>O-5C3m(l\3 x2 = 0.0001) and in 1995 (Pr > x2 = 0.0001) compared to control plots (Tables 19). Treatment differences for the 6-year burn plots were significantly different from controls only in 1994 (Pr > 12 = 0.026). Traps in 2-year and 6-year treatments in 1994 were 2.7 and 1.5 times more 59 0.0000 30.. 000. .0 0.02000 ..:..03 .0: 0.0 058033.. .3. .0. 0.00 0... .0 000:0.030. 0.0300 30.. 30. .0 0.03000 ..E.03 .0: 0.0 3.03 .0 .0. 0.00 0... .0 0005030.. 0.5.00 30.. 30. .0 0.02000 .....:03 .0: 0.0 0000.00 .0 .0. 0.00 0... .0 00000030m 0.0300 30.. 000. .0 0.00000 ..E.03 .0: 0.0 0.3.3200 .0 .0. 0.00 0... .0 00000030.. .8... v 3. .000... 0... 0. 00.50.0000 E00530 0 00.. .0... 00.00..0> .0. 2:0 00:030. 0.0 02.0. 00000 A... A .3. 00 00:030. .0 u 8.00 - 02.00 .0... 0.00. 0.00509... .0 0.3003N .5 3.83 85 8:08 08.: 30.. ___2,_ 0.020 .0... 3500023 0... .0 0.00.500... .0..:00 00:50:: 0... :0... 002000 0.0:. 00.00 :5. 0.23 E00 .00>-N .0... 3500023 0... .0.. ”.00 0.00 0... :. 0_00..0> :0..00_.._000_0 00.020 .00_ 00. :0... 002000 0.0:. 00.00 :5. 0.020 .0 60:000.. 00.000030. 0... .0... 3500023 0... 2000.30. 02.0. 0000.. .0... .0... S8... 00.0 0.0 08... 8... 8... .8... 0.. 8.0 .0 .80... 800.00 0.8 8.0 8... S... I... 0.... i E .3803? K 00... 8... 0.... 0.... 8... S... .8... mm... 8.0 00.0 0.80. .3 0.... 00... 08.0 8.0 0.. 8.0 8... 8.0 0.... 8.0 0.. 8.0 08.300 .3 0.... fl... 8... 0.... 0.0 8... 8.0 R... 8... 8.0 0.0 8.0 8.2.: .3 8... N8... 8... 0.0 .8... 5. . .8... 3.003 .0 8.0 8... S... E... 4.0 8... .0028 .0 8... .0... 8... .02... n--- 3... .8858 .0 2.0. 2.0. 2.0. .0..0. N 0.8... :8... 0 .9, N 0000 .00. 0 0000 .00. N N 3.02.. IS... .0 .9 N 0000 .00. 0 0000 .8. a 00.0030 08. .8. .0.0>.0:0 0.00 .00..030.00 .00.. 00.00..0> .:00:0300:. .0. 00.0.. 0.0 02.0. 0000 0:0 A «x A .3 . 00:_0> >.._.0000.3 .0:00000 32330.. 003 0:0 000.. :. 00.00 0.00.00 .0.0. 0:0 00.0030 0.00.00 00330.. 2:08:80 .00:. :0>00 0.... .0 0.:000 30.. 80.. 0.300. 0.0205 :200050. 2.030.. .2. 0.00... 60 likely to catch more beetles than control treatments. Traps in 2-year burn plots in 1995 were 2.5 times more likely to catch beetles than the control. Differences between blocks proved significant in 1994, but not in 1995. Blocks l and ii, with totals of 291 and 296 respectively, were not different from each other, but were significantly different from the 385 beetles trapped in block Ill (Pr > )8 =0.041 and 0.035, respectively) (Table 20). In both years, hypothesis tests indicate that differences between the 2-year and 6-year burn treatments were highly significant (Pr > x2 = 0.001 ). Not all Carabid species reacted to burn treatments with this pattern. Burn treatment effects were not significant for Trichotichus vulpeculus. Pterostichus tristis and C. convivus did not show significant differences in activity among treatments; in 1994 for C. convivus, both years for P. tristis. The effect of blocking was significant for P.tristis; this species was twice as likely to be trapped in block l as in block lll (Table 20). Data for C. convivus were too sparse in the 6-year—burn treatment to allow analysis for 1995, even though the 47 beetles trapped of that species represented 6.8% of the total season catch. A species closely related to C. convivus, C. soldalis, showed significant activity differences in 1995 between 2-year burn and unburned treatments(Pr > 12 = 0.02). Six-year burn treatment activity was not significant from control levels (Pr > x2 = 0.74). Burn treatments were significantly different from each other (Pr > x2 = 0.01). 61 Table 20. Block totals for nine common Carabid species and total number of beetles trapped in 1994 and 1995. 1994 1 995 Species Block I Block II Block "I Block I Block ll Block Ill C.convivus 12 23 26 14 14 19 C. soldalis 8 22 7 11 20 16 C. goryi 4 27 28 1 32 29 P. mutus 23 15 37 17 16 15 P. stygicus 135 88 120 84 48 49 P. tristis 5 17 26 6 14 1 5 S. stenostomus lecontei 18 1 1 15 6 5 7 S. impunctatus 14 47 78 2 7 17 T. vulpeculus 19 1 1 1 31 14 1 1 Total Carabids 291 296 385 241 222 232 62 One or more burn treatments had significant effects on C. goryi, P. stygicus, and P. mutus. Traps in 2—year and 6—year burn treatments were 11 and 9 times more likely to catch more C. goryi than control treatments (Pr > x2 = 0.001 and 0.004), respectively. Six—year burn plot activity was not significantly different from that on 2-year plots (Pr > x2 = 0.56), and significant differences occurred between blocks l and III (4 beetles compared to 28). Pterostichus stygicus had significantly greater activity in 1994 and 1995 on 2-year burn plots than unburned (Pr > x2 = 0.06 and 0.02, respectively), but not on 6-year burn plots (Pr > 12 = 0.70 and 0.36, respectively). Differences between the two burn treatments were significant in 1994 (Pr > x2 = 0.002), but were not in 1995 (Pr > X2 = 0.12). Significant differences between blocks occurred only in 1994 for blocks ll and Ill (Pr > x2 = 0.02). Pterostichus mutus also showed significant treatment effects for the 2- year treatment (Pr > x2 = 0.005 and 0.02), but not the 6-year burn (Pr > x2 = 0.77 and 0.79) in both years. Two year burns were significantly different from 6-year burn plots (Pr > X2 = 0.01 and 0.03). Differences between blocks were not significant. The observation of greater Carabid activity on burned plots vs. unburned treatments in my study is corroborated by a few studies in grasslands and forests. Hansen (1986) collected greater numbers of insects following a 20,000 ha fire on shrub-dominated range sites than on unburned sites. After severe 63 wildfire in Alaskan black spruce stands, total arthropod abundance was higher on burned than similar unburned stand areas (Viereck and Dyrness 1979), although the Coleoptera component of that community was less abundant during the first post-fire year relative to pre-burn activity levels. This decrease was attributed to greater predation success by spiders on burned areas. In contrast, other studies showed that Carabid activity on burned forest sites is reduced following fire. Holliday (1991) observed that most Carabid species become locally extinct during intense forest fires in northern boreal coniferous forests, or during the extreme habitat changes that follow. In a 10- year study of the same site, Carabid species abundance was generally 34% of unburned control plots (Holliday 1992). Harris and Whitcomb (1971) reported higher abundance among four species of Evarthrus (Carabidae) in unburned compared to burned southern pine stands. In a later study, Harris and Whitcomb (1974) noted that annually burned southern pine plantation sites supported consistently lower Carabid beetle populations than sites that had been protected for 10 years. Other researchers noted a 60 percent decrease in beetle populations in southern pine stands following prescribed burning (Pearse 1943, Heyward and Tissot 1936). Ahlgren (1974) found fewer beetles in prescribed burned jack pine stands in Minnesota. Repeated controlled burning had little effect on beetle abundance in other studies. Researchers in Australia found that two consecutive prescribed fires at a 3-year interval had little effect on total arthropod abundance, although activity 64 of Collembola declined for one year following each fire (Collet et al. 1993). Rickard (1969) found greater abundance of phytophagous beetles on unburned shrub steppe compared to burned sites, but concluded that fire had little effect on the only Carabid studied, Calosoma luxatum Say. On sandhill sites in Florida, differences in abundance of common ground dwelling beetle species (Coleoptera: Carabidae and Scarabidae) among plots burned at 1, 2, and 7 year intervals were not significant from unburned sites, although 5-year interval plots were (McCoy 1987). Correlation analysis indicated that the abundance of the four most common beetle species were positively related to time since burning. Observations of greater beetle activity on 2-year compared to 6-year burn treatments in my study is consistent with the findings of most fire studies in grasslands, but not forest sites. Spider populations in annually burned prairie grass-dominated wetlands were significantly more abundant than populations in unburned sites, consistent with the higher vegetative productivity observed following burning (Johnson 1995). Riechert and Reeder (1972) suggested that spider populations in dry prairie sites in Wisconsin are well adapted to fire, noting that periodically burned areas supported stable populations. The results of these studies may not be directly applicable to forested ecosystems, however. Gillon (1972) observed that populations of sun-loving Pentatomidae were maintained by periodic fires in African savannas. Metz and Farrier (1973) found that annually burned plots in North Carolina loblolly stands contained significantly lower soil mesofauna abundance than unburned and periodically 65 burned plots. They noted that soil mesofauna recovered to unburned stand population levels after 38 to 46 months, however. In a similar study, Metz and Dindal (1975) observed significantly lower abundance of Collembola in annually burned loblolly stands compared to unburned sites, but no significant differences between periodically burned (every 5 to 8 years) plots and controls. Vegetation Characteristics of Beetle Activity The relationship between vegetation characteristics and beetle activity was examined by fitting regression models through categorical data analysis. This procedure used only 1995 ground flora and beetle count data and permitted analysis only for total carabids and five of the common beetle species (Table 21). Total Carabid Activity Models--When fitted without treatment variables, the 1994 total Carabid activity model incorporating overstory basal area, sapling, and seedling density variables showed significant differences only for sapling density (Pr > x2 = 0.0003). Both variables were negatively related to beetle counts, with odds ratios of 0.98 for seedlings and 0.86 for saplings. Using 1995 total Carabid data only, seedling (Pr > X2 = 0.03) and season variables were significant (Pr > x2 = 0.0001). Again, seedling density was negatively related to Carabid catches, with high density plots 0.98 times as likely to catch beetles as low-density plots. This relationship confirms the overall results of pitfall trapping in both years of the study. More beetles were trapped on the 2-year burn plots than on unburned plots in 1994 and 1995, possibly reflecting the almost complete lack of seedling and sapling size vegetation on these frequently 66 Table 21. Summary of significant variables in logistic regression models relating vegetation characteristics to beetle activity. Variables included in each model are indicated by entries of NS for not significant, or Pr > X2- Species 2- 6- Over- Sapsz Seed3 Herb“ Wood“ Spb Su’ year year story1 burn burns C. soldalis 0.005 0.04 0.05 0.03 0.02 NS NS -- -- P. tristis NS NS NS NS NS NS NS -- -- T. vulpeculuse -- -- NS NS 0.04 NS NS -- -- P. mutus NS NS NS NS NS NS NS -- -- P. stygicus9 NS NS NS NS NS NS 0.004 -- -- P. stygicus10 -- -- NS 0.09 NS NS 0.007 -- -- Total -- -- NS 0.001 NS -- -- -— -- Carabids11 Total -- -- NS NS 0.03 -- -- 0.0001 0.0001 Carabids11 Total NS NS NS 0.05 NS -- -- -- -- Carabids12 Total 0.004 NS NS NS NS -- -- -- -- Carabids13 Total NS NS NS NS 0.07 NS NS -- -- Carabids” Total -- -- NS 0.15 0.15 0.09 NS NS 0.04 Carabids14 Total -- -- NS NS 0.02 NS NS NS 0.02 Carabids15 WW = basal area at each trap location for trees >5.9 cm dbh. 2Saplings = stems per ha 2.0-5.9 cm dbh. 3Seedlings = stems per ha at least 1 m tall and 0-1.9 cm dbh. 4Herbs = percent ground coverage of herbaceous ground flora in 1 m2 quadrats. 5Woody = percent ground coverage of woody ground flora < 1 m in 1 m2 quadrats. 6Sp = Spring season, or trapping periods 1, 2, and 3. 7Su = Summer season, or trapping periods 4, 5,6, 7, and 8 8Only the model fitted without treatment variables attained significance; this model had poor fit. 9This model also included variables for seasons; only spring was significant (Pr >38 = 0.001). 1OSeason variables in the model were not significant. Treatment variables were not included. 11These models used data for 1994 and 1995. 12Model used 1994 data. 13Model used 1995 data, but did not include ground flora variables. 1“Model used 1995 data only. 15Model used 1995 data only and included block variables; only block 2 was significant (Pr > X2 = 0.09). 67 disturbed sites. With the addition of treatment variables, the 1994 model showed significance only for saplings (Pr > x2 = 0.052); saplings remained negatively related to beetle activity. Treatment variables were not significant in the combined model (Pr > x2 = 0.19 and 0.86 for 2- and 6-year treatments). Seedling density became insignificant for 1995 with the addition of treatment variables (Pr > 12 = 0.22). The 1995 model did show a significant 2-year burn treatment effect (Pr > x2 = 0.004) but not a 6-year effect (Pr > x2 = 0.81 ). The addition of ground flora variables to 1995 total Carabid activity models adds little predictive power. Herbaceous ground flora coverage and woody ground flora effects were insignificant. Seedling density had better significance than in the model without ground flora variables (Pr > x2 = 0.07) and became weakly positively related to beetle catches. Herbaceous ground flora coverage has greater significance (Pr > )(2 = 0.09) when treatment variables are removed from the model, while sapling and seedling variables both approach Pr > x2 = 0.15. Herbaceous ground flora, seedling, and sapling variables have very low estimates for the regression equations; all three were less than 0.01, translating to odds ratios between 1.0 and .99 for chances of higher catches than low density or coverage plots. Ground flora variables did not have a significant contribution even when run in models with block and season variables (Pr > x2 = 0.39 and 0.26). Thus, it is possible to say that ground flora differences among treatments are not significantly related to beetle activity when the 68 variance due to seasons, blocks, saplings, seedlings, and overstory density are removed. C. soldalis models--Seedling and sapling density were significant (Pr > 12 = 0.02 and 0.03) in models incorporating treatments, blocks, dates, and ground flora variables. Overstory basal area differences were marginally significant (Pr = 0.053), perhaps reflecting a relationship between crown closure and C. soldalis catches. Burn treatment effects were also significant (Pr > 12 = 0.005 and 0.04). Sapling, seedling, and overstory basal area variables were directly related to catches, with odds of higher catches on high-density plots‘of 1.9, 1.1, and 1.05 times the odds on lower density plots, respectively. Seedling and sapling density were not significant in models fitted without treatment variables, whereas basal area differences among traps remained significant (Pr > x2 = 0.01). Ground flora variables were not significant in predicting catches (Pr > x2 = 0.96 and 0.58). P. tristis models--A|l vegetation variables were insignificant at a = 0.05 in models for P. tn'stis activity, even when they were fitted separately to the beetle data. T. vulpeculus models-Seedling density was inversely related to catches (Pr > x2 = 0.04) in the model without treatment variables. Ground flora and other vegetation variables were not significant. This model had very poor fit according to the Pearson residual statistic, however (Pr > X2 = 0.0025). 69 P. mutus models--Herbaceous ground flora showed a slight but non- significant relationship to P. mutus activity (Pr > x2 = 0.14) in models testing solely the effect of ground flora. Models including other variables were not significant. Herbaceous coverage was only weakly and inversely related to higher beetle catches; the probability of catching more beetles on plots with greater coverage were only 0.97. P. stygicus models--Woody ground flora and seedling density were significantly related to P. stygicus activity in models with treatment, block, season, overstory, sapling, seedling, and ground flora variables. Woody ground flora was weakly, but directly, related to catches (Pr > x2 = 0.004), while seedling density was not significant (Pr > x2 = 0.12). Seedling density was also directly related to beetle activity. These variables had odds of higher catches of 1.02 and 1.04, respectively, on higher density plots. In models without treatment variables, woody ground flora remained significant (Pr > X2 = 0.01), although the attendant odds ratio did not improve. Sapling density was directly related to P. stygicus activity (Pr > X2 = 0.09), albeit with a low odds ratio of 1.2. This merely reflects treatment effects on sapling density. Seedling density became insignificant with the removal of treatment variables. Although some vegetation variables appear to be significant in explaining variation in beetle activity, most of the odds ratios are very low, indicating little predictive strength. At the species level, C. soldalis was most strongly associated with any vegetation variable. The odds ratio for overstory stocking of 70 1.93 suggests that microsite variations in basal area may have a significant influence on activity for that species. Sapling density has little effect; the odds ratio for higher catches is nearly zero. The relationships between the vegetation density or coverage and activity for the individual species tested appear to contradict species totals for the treatments. In this study, this is probably the result of species specific habitat preferences, and sparse data for each species. Twice as many C. soldalis were trapped on 2-year burn plots than on the other treatments, yet the regression model predicts greatest activity for this species on sites with higher overstory and sapling density. Estimates for total beetle activity suggest an inverse relationship between herbaceous vegetation and beetle activity; yet nearly twice as many Carabids were trapped on the 2-year burn treatments, which had the highest average and total herbaceous ground flora coverage. This disparity may be partly due to low beetle catches during each trapping period. SAS Logistic procedures require that a majority of the observations in each data class have counts greater than 5 for best performance, yet this condition was not easily met in my study. Regression models for P. stygicus suggest a positive relationship with sapling, seedling and woody ground flora density. The largest number of this individuals of this species were trapped on 2-year burn plots, where herbaceous coverage was highest, and other vegetation variables lowest. Six-year burn plots had the highest average coverage by woody ground flora, but the lowest beetle activity. Simple examination of the data indicate that slightly more than 71 half of the P. stygicus were trapped on plots with woody ground flora coverage greater than 10 percent. Although most of the P. stygicus were trapped on 2- year burn plots, 61 beetles were caught in control plots, hence the relationship with seedling and sapling density. Several explanations are possible. For total Carabid data, the predicted relationship may reflect the habitat preferences of the most active species trapped: P. stygicus, T. vulpeculus, C. soldalis, and P. mutus. Regression models for these species all indicated negative relationships between activity and herbaceous vegetation coverage, or stronger relationships with woody vegetation components. Logically, a great deal of variation in microsite preference should exist for the 45 species trapped. This variation may thwart attempts to associate activity with one or a few common vegetation variables. While P. stygicus is described as characteristic of moist forest sites in Lindroth's (1969) classification, it is possible that the species is fairly insensitive to variation in vegetation characteristics, given a uniform overstory. The crudely 3-dimensional measurement of vegetation coverage may not adequately reflect important habitat characteristics. Habitat suitability for insects is clearly affected by vegetation structure. The availability of suitable microsites for habitation by herbivores may be positively related to floristic heterogeneity, patchiness of vegetation, and plant architecture (Southwood et al. 1979; Schowalter 1986; Lawton 1983; Lawton and Schroder 1977) or may simply be correlated with other relevant habitat variables 72 influencing species abundance. Murdoch (1972) found that Homopteran species diversity was correlated with plant structure and diversity, although others have found no significant correlation with vegetation parameters (Hawkins and Cross 1982). Changes in availability of cover and density of vegetation following fire may provide protection from predation for herbivores and Carabids from predators. McCoy (1987) observed that entomophagous reptile and amphibian abundance was negatively correlated with ground cover density in periodically burned sandhill sites and unburned areas. Abundance of four species of Carabid beetles common to sandhill sites was positively related to density of vegetation cover (>10 cm) and vegetation density at ground level (<10 cm) in the same study. At least one of these four species is known to be consumed by lizards on the site from studies of gut contents. Drastic reductions in vegetation cover and density following more severe burns may expose invertebrates to greater rates of predation (Anderson et al., 1989). Little is known about the rates or factors affecting predation on ground beetles. Some arthropod species activity may be related to responses of certain key plant species that are favored by fire. Hansen (1986) observed higher abundance of leafhoppers on burned shrub-dominated range sites compared to unburned sites, possibly related to re-growth of preferred browse species. Paradoxically, parasitic wasps including ichneumonids, brachonids, and mutillids were attracted to burn sites in the same study, despite an absence of flower food sources, and presumably, a lack of potential hosts. Changes in the abundance of grass- and forb-eating grasshopper species in burned and unburned prairie 73 sites was related to shifts in the dominance of grass species on prescribed burned areas (Evans 1984, Kucera and Koelling 1964). Other species may be attracted to charred wood and debris created by fire. Richardson and Holliday (1982) noted that P. adstn'ctus occurred most frequently on sites with greater amounts of fallen trees and rotting logs, characteristics most common on burned spruce sites. Activity of the most abundant species may simply reflect changes in prey density, or changes in nutritional quality of vegetation on burned sites. Holmes et al. (1993) related Carabid beetle distribution to environmental variables including nutrient status, saturation of substrates, altitude, and grazing in a large scale survey of peatlands in England. Higher foliar concentrations of Nitrogen may temporarily result following litter reduction during light burns. Foliar N concentration was positively related to phytophagous insect survival in Minnesota jack pine stands (McCullough and Kulman 1991). A short-term fertilization effect following light burning may have favored higher phytophagous prey populations; yet more severe fires are likely to reduce N availability through volatilization and soil leaching on sandy sites. Light fires may also speed litter breakdown or make conditions more favorable for saprophages and mycophages by virtue of its spotty coverage and partial consumption of large debris. Removal of the dry needle litter and partial consumption of the fermentation layer may result in rapid breakdown of remaining substrates due to greater insolation and surface temperatures. Metz and Dindal (1975) reported that the greatest number of Collembola occurred in the F and H layers of 74 unburned and periodically burned (5- to 8- year intervals) southern pine stands. In turn, this may attract Carabids. Activity is unlikely to be influenced by changes in the abundance of any single prey species, however. Although Carabids have generally been considered predatory, several authors have suggested phytophagy under certain conditions for many species (Johnson and Cameron 1969, Hengeveld 1980). Loreau (1982) suggested strong polyphagy among small (less than 6 mm) and large (>15 mm) Carabid species and a high degree of adaptability in dietary habits for almost all beetles. Most Carabids are prey on primary consumers, chiefly on myco- and saprophagous species. These include Oligocheta (Lumbricidae and Enchytraeidae), snails and slugs, oribatid mites, Collembola, and nematoceran Diptera (Loreau 1982). Carabids tend to be opportunistic in their dietary habits; Loreau reported that single species formed no more than 15% of any beetle’s diet in laboratory studies. Only the medium-sized (6-10 mm) beetles in Loreau’s study exhibited a high degree of specialization on Collembola and mites. Larger beetles (>10 mm) formed the majority of the Carabids trapped in this experiment. Structure and density of vegetation layers may also affect the favorability of habitat microclimate for invertebrates. Thiele (1977) suggested that Carabids are quite sensitive to differences in microclimate between sites. Vegetation structure and density may moderate humidity and temperature extremes in microsites (Gieger 1966, Richardson and Holliday 1982). Neve (1994) found Carabus spp. activity positively correlated with mean daily temperature in 75 European beech-dominated forests. Activity was most closely related to maximum daily temperatures and day/night contrasts in temperature. Humidity also appeared to be positively related to activity (Thiele 1977, Neve 1994) but has been discarded as a variable in other studies (Desender 1983). The complex interaction of edaphic, microclimate, and vegetation factors make the isolation of specific variables responsible for species distribution difficult (Richardson and Holliday 1982, Holmes et al. 1993). Species Diversity Prescribed burning clearly led to increased beetle species diversity on burned plots relative to unburned areas in my study. Burn treatments consistently had greater species richness than control treatments, according to six common diversity indices and formal model fitting (Table 22). Measures differed on ranking the diversity of the two burn treatments, however, according to the diversity element each reflects. Overall, pitfall trapping produced a total of 45 species over the two years of the study (Table 17). Thirty species were collected in 1994, 39 in 1995. More locally ‘rare’ species were collected in 1995; 10 species were represented by one individual, whereas single individuals of only four species were collected in 1994. All of the species captured have relatively wide geographic ranges, and are relatively common in the United States, according to Lindroth (1969). Catches also included three “introduced" species from Europe; Amara familiaris Duftschmid, Harpa/us puncticepts (Stephens), and Pterostichus melanarius (llliger). One species trapped, Amara impunctico/lis (Say), has a holarctic range. 76 Table 22. Diversity indices by year for burned and unburned treatments. Diversity 1994 1995 Measure Control- 6-year 2-year Control- 6-year 2-year no burns no burns Species richness (S) 18 25 27 24 27 35 No. of Individuals 233 295 444 177 140 378 Maraglef 3.2 4.2 4.3 4.4 5.3 5.7 Berger-Parker (N) 2.1 3.1 3.2 2.9 4.4 4.3 Shannon (H') 1.9 2.3 2.4 2.4 2.8 2.7 Shannon 0.68 0.72 0.72 0.76 0.84 0.76 evenness (E) log series (a) 4.6 6.5 6.3 7.5 9.9 9.4 log series or from 5.5 6.0 6.5 8.0 9.5 9.0 William's nomograph log normal (9.) 34.9 50.2 50.1 58.2 68.2 73.6 77 Burn treatments consistently had greater species richness than unburned plots, as measured by Maraglef’s index, which measures the number of species present in a community relative to the total number of individuals sampled. Two- year burn plots were more diverse than 6-year burn plots. This index is highly sensitive to sampling bias, however (Magurran 1988). The Berger-Parker index (D) shows a higher degree of dominance by the most common species in the control treatments in both years relative to burn treatments. This index also shows higher beetle diversity on burn treatments. Six-year burn treatments had higher species dominance and lower diversity than 2-year burn treatments in 1994; however, the opposite prevailed in 1995. The Berger-Parker index is relatively independent of sample size; however, it can be biased by fluctuations in the abundance of the most common species and this is most likely the case with 1995 six-year burn data. Species abundance indices also show greater diversity on burned relative to unburned plots, but differ on the relative diversity between burn treatments. Calculations of the log series a indicate greater diversity on 6-year burn treatments in both years, a reflection of the greater number of species with intermediate abundance in that treatment. The log normal A and Shannon diversity (H’) statistics tend to reflect more closely the number of rare species. These measures rated diversity on 6-year burn plots higher than 2-year burn plots in 1994, but lower in 1995. The Shannon evenness (E) statistic indicates higher diversity on burned compared to unburned treatments due to the equability of species abundance in 78 both years. Differences in burn treatment evenness followed the same pattern described by the log normal it and Shannon diversity indices. Pair-wise comparisons of Shannon diversity (H’) estimates were conducted using t-tests (Table 23). Differences between estimates for burn treatments and control treatments were highly significant (Pr > .001) in both years. Diversity estimates were not significantly different between burn treatments, however. Given the similarity of diversity index measurements between 2-year and 6-year burn plots, the safest conclusion would be to reject only part of the third null hypothesis; while burning clearly increased beetle diversity in this study, the diversity measures employed give contradictory results when ranking diversity between frequently and periodically burned treatments. I believe this discrepancy reflects sensitivity of the indexes to small sample sizes on the 6- year burn plots in 1995 and changes in the abundance of the most commonly trapped species, P. stygicus. This bias probably obscures diversity patterns between the two treatments; 2-year burn plots contained more rare species and greater absolute species richness than plots burned every six years. In light of these observations, I conclude that Carabid diversity does increase with the frequency of prescribed burning. The log series had the best fit of the four common species abundance distribution models for all three treatments in 1994 and 1995 (Pr > X2 > 0.05) (Table 24). The log normal series also fit treatments well, with the exception of control plots in 1995. Visual inspection of the predicted and observed species 79 Table 23. T-tests1 for Shannon diversity index estimates by year and burning treatment. 1994 1995 Shannon Control- 6-year 2-year Control- 6-year 2-year T-Tests no burns no burns Shannon (H') 1.9 2.3 2.4 2.4 2.8 2.7 variance 0.007 0.005 0.003 0.007 0.007 0.003 t-statistic 3.09 0.69 4.04 2.92 0.45 2.96 df 492 602 433 314 277 337 2 NS 3 2 NS 2 A B B A B B 1means followed by the same letter are not significant at a = 0.05; 2significant at alpha = 0.01 3significant at alpha = 0.001 Table 24. Goodness-of—fit summary (a = 0.05) for four common species abundance models. 1994 1995 Fit of Models Control 6-year 2-year Control- 6-year 2-year no burns no burns Geometric no no no no yes no Log series yes yes yes yes yes yes Truncated Log yes yes yes no yes yes Normal Broken Stick no no no no yes no 80 abundance shows that the log normal series predicts more intermediate and highly abundant species than observed on unburned plots in 1995 (Pr > x2 > 0.05). Data for the 1995 six-year burn plots fit all four models adequately; this is undoubtedly due to small sample bias (n = 140). The geometric series and broken stick model did not fit the data well. A good fit for the log series model suggests that relatively few factors dominate the ecology of these communities (Magurran, 1988). These factors could include the frequency of prescribed burning and the amount of sunlight reaching the forest floor. The abundance patterns typical of the log series model have also been interpreted to reflect the occupation of niche hyperspace at random intervals by species migrating into unsaturated habitats (Boswell and Patil 1971, May 1975). This model is generally associated with communities in the early to intermediate stages of succession; those with relatively few species and limiting environmental factors. Although the log normal model also fit most treatments well, the abundance distribution predicted did not differ significantly from the log series model (Pr > X2 > 0.05) for all but the 1995 6-year burn plots. Six-year burn plot estimates in 1995 differed chiefly in the number of species predicted to have the highest abundance. Magurran (1988) recommended that the log normal series be fitted only for continuous data, although it is frequently used for abundance. The observation of generally higher diversity on frequently burned plots over unburned treatments in my study is consistent with the results of few other studies on prescribed burning. Metz and Dindal (1975) reported increased 81 diversity of Collembola species on periodically burned plots as calculated by a modified Shannon-Wiener index. Annually burned plots exhibited decreased Collembolan diversity in the L layer of the forest floor but increased diversity in the F + H layers, compared to control plots. Other researchers found that controlled burns produce little change in insect diversity. Relative to pre-burn levels, consecutive prescribed burns in Australian eucalypt forests produced no significant differences in insect taxa species diversity, richness, and evenness following the first burn (Collet et al. 1993). Species diversity was higher following a second burn conducted 3 years later, primarily due to an increase in species evenness, rather than an increase in species richness. McCoy (1987) noted little difference in Carabid species richness on periodically burned and unburned sandhill plots in Florida. In wildfire studies, several researchers have reported no significant effects on species diversity, evenness, and, in some cases, abundance of Carabid communities following intense burns. Holliday (1992) noted no differences between species evenness and species diversity as measured by at in mature stands and burned sites in two boreal forest types over a 10-year period following wildfire. Greenberg and Thomas (1995) reported no significant differences between diversity, evenness and abundance of Coleopterans in a study comparing effects of high intensity fire and salvage logging to other harvesting disturbances and undisturbed forest sites in Florida sand pine scrub, 5 to 7 years after disturbance. By contrast, Winter (1980) noted greater 82 numbers of species and individuals in young pine stands for 1 to 3 years after wildfire in Germany. Arthropods inhabiting non-forested ecosystems exhibit varied responses to prescribed burning. Anderson et al. (1989) noted that grasshopper abundance and diversity were unaffected by burning in prairie grass ecosystems, while abundance of other insects were depressed for one year, recovering rapidly to pre—burn levels thereafter. Evans (1988) found that species richness of grasshoppers was greater on infrequently burned or unburned sites than annually or biennially burned prairie grass sites. In addition, grasshopper species richness was found to be positively correlated with local plant species richness and diversity. Book and Bock (1991) observed a one-year decrease in grasshopper abundance following wildfire in an Arizona grassland. This trend was attributed to direct fire-caused mortality and a reduction in grass cover during the first year following the fire. Johnson (1995) reported greater spider densities and species richness in annually burned vs. unburned wetlands dominated by prairie cordgrass (Spartina pectina). Cessation of burning in restored tall grass prairie in Wisconsin caused a reduction in soil arthropod diversity and abundance, however (Lussenhop 1976). Lussenhop suggested that the decline may have been due to reduced root biomass productivity and detritus following cessation of burning. Beetle diversity changes following prescribed burning may be linked closely to changes in vegetation diversity. Parallel patterns of diversity following disturbance have been observed between insect fauna and flora communities. 83 Southwood et al. (1979) and Brown and Southwood (1987) noted similar trends in macrofaunal diversity and floral diversity during old-field succession to deciduous forest. Species abundance and richness increased with successional age, while diversity as measured by a rose for one year, then remained stable. The authors attributed initial increases in invertebrate diversity to increasing plant taxonomic diversity, while subsequent stabilization was attributed to increasing spatial diversity of flora, which compensated for the effects of declining plant taxonomic diversity. During colonization of mining spoils by forests, researchers noted a weak trend toward increasing Carabid species abundance and diversity with increasing successional age in one study (Neumann 1971), but not in two others (Mader 1986, Hejkal 1985). Harvesting disturbances appear to produce opposing patterns of Carabid diversity. Increased diversity of Carabid species assemblages on clear-cut relative to uncut forest sites was noted by several researchers (Niemala et al. 1993, 1990; Jennings et al. 1986). Lenski (1982) reported a significant increase in within-genus species diversity after clearcutting in Appalachian oak-hickory forests. Carabid beetle populations showed greater overall abundance and species richness than undisturbed mature stands in clearcut boreal pine-spruce forests in Canada (Niemala et al. 1993). By contrast, experiments with strip clearcutting in dense spruce-fir stands in Maine showed lower abundance in clearcut strips and uncut dense stands compared to uncut residual strips (Jennings et al. 1986). Species diversity was greater in strip clearcut stands than in dense undisturbed stands, however. Studies in dry site pine stands in 84 Poland also found greater diversity and dominance in regenerating stands (Szyszko 1987). Several theories regarding arthropod responses to community disturbance have been advanced to explain changes in species diversity. According to interspecific competition theory, species diversity should decrease in the absence of disturbance, as competitively superior species exclude inferior competitors (Dayton 1971; Sousa 1979). Lenski (1982) and Loreau (1986) consider interspecific competition to be a driving force in the structuring of Carabid beetle assemblages. A contrasting theory advanced by Liebherr and Mahar (1979) contends that arthropod diversity and abundance are more closely related to vegetation structure and habitat complexity or heterogeneity than to the frequency and magnitude of disturbance. They reported greater beetle diversity and higher abundance in more successionally advanced, structurally diverse stands during a survey in Michigan dry-site oak forests. Species dominance decreased with canopy closure, increased understory development, and floristic diversity. Insect species diversity may be influenced by plant taxa diversity, as well as spatial and architectural complexity in a habitat (Southwood et al. 1979, Schowalter 1986; Lawton 1983). Important mechanisms governing insect diversity responses to disturbance may include interspecific competition, size of the potential colonist pool for given sites, and the rate of change in plant species diversity following fire. Holliday (1992) reported estimates of species gain and loss for burned and 85 unburned boreal forest sites in Manitoba following an intense forest fire. Lower beetle abundance on burned or early successional stage sites throughout the experiment period were attributed to lower rates of species gain on burned sites. Burned and unburned plots appeared to have similar rates of species loss. Unfortunately, little is known regarding Carabid species colonization and extirpation rates in temperate forest regions. Qualitative aspects of species diversity The results of my study do not indicate any dramatic changes in species gain or loss, nor can they be estimated given only two years sampling data. Several new species to the site were encountered in 1995 compared to 1994 collections, but it is not clear whether this represents colonization or merely sampling variability. Most of the species I captured are typical of dry, sandy environs, and about one third are typical of forests, according to Lindroth’s classification (1969). Ten of the species collected in the Tribe Harpalini are customarily found in open, grassy meadow habitats, usually on dry sandy soils. The three Amara species collected have been observed by other researchers on dry sandy sites. Cymindis americanus is an almost exclusively xerophilous species, occurring on open sandy country with sparse vegetation, but also among dead leaves at the edges of forests. Most of the Harpalus spp. sampled were trapped on burn treatments. Habitat preferences of 10 other species included mesic deciduous forests; Agonum melanarium, A. ferreum, Chlaenius tricolor tricolor,and Platynus hypo/ithos were most typical of habitats near streams or water. 86 Dicaelus e/ongatus has been found in both dry hillsides and among moist leaves in swamps. Fourteen of the species trapped are typical of forests, only 7 of which inhabit dry woodlands. Unfortunately, little is known about the variation in habitat preferences for these species; descriptions of the ecology for most species is anecdotal. Lindroth’s classification (1969) often lists several microsites for each species, sometimes contradictory. The four most abundant species on control plots included Carabus goryi, Pterostichus stygicus, Synuchus impunctatus, and Platynus hypo/ithos, all species typically found in mesic forests or moist microsites. Pterostichus stygicus, S. impunctatus, and C. goryi are also among the most abundant species on 2-year burn plots, along with P. mutus, a species typical of xeric open sites. Although diet habits of Carabids are not well known, most of the species trapped are considered carnivorous, although some tribes of Carabidae are known to be polyphagous. Three large-bodied species, Dicaelus elongatus, Sphaeroderus stenostomus lecontei, and Carabus goryi, are known to prey on large insects, including snails and caterpillars. More 8. stenostomus lecontei were trapped on control and 6-year burn than on 2-year plots, while C. goryi was caught most frequently on burn treatments. Members of the Notophilious, Agonum, and Ga/erita genera are also decidedly carnivorous. Under specific conditions, even predaceous genera, including Carabus and Pterosticus, have been known to search for food and water supply in fresh or rotten fruits, or in green parts of plants (Lindroth 1949, Johnson and Cameron 1969). Johnson and Cameron (1969) observed phytophagous behavior among 87 generally predaceous members of the Pterostichus and Agonum genera. Species of the Harpalini have generally been considered to have herbivorous or opportunistic predaceous food habits; these were found most frequently on burn plots where herbaceous ground floral coverage was greater. Harpalus species are known to alternate from predation to seed eating (Brandmayr 1990). True spermophagus Carabids, belonging to the Amarini, Zabrini, and Harpalini tribes, generally subsist from the small fruits of herbaceous plants, such as are found in the Graminaceae, Umbelliferae, Cruciferae, Labiatae, and Compositae (Brandmayr 1990). Some species, including Synchus impunctatus Say (Manley 1971) and Harpalus pennsylvanicus (Kirk 1972) are known to store seeds in burrows as adults or larvae. In diet studies, Harpalus honestus larvae were reported to exhibit rapid development solely on seeds from Taraxacum spp. (Brandmayr 1990). Therefore it seems likely that greater herbaceous plant community development on 2-year burn plots in my study might be related to the observations of higher beetle activity. Chapter 4 CONCLUSIONS AND SUMMARY The work reported in my thesis tests a global operating hypothesis that the responses of an insect community, as represented by Carabid beetle fauna, to controlled, low intensity fire can be explained more completely by the fire- induced vegetation characteristics of the site, rather than by fire frequency. This broad concept was examined by focusing on three corollary hypotheses: 1. Total Carabid activity increases with frequency of burning, and diminishes with time since disturbance. 2. Carabid activity is negatively related correlated with the density of woody understory vegetation (saplings and seedlings), but is unrelated to ground flora coverage. 3. Carabid species diversity decreases with the frequency of fire, and increases with the time since disturbance. Acceptance or rejection of the appropriate null hypothesis for these three theories could provide insight into the causal relationship between insect fauna response to periodic disturbance. Results of this experiment suggest that Carabid beetle activity, as evidenced by their abundance in traps, does increase with the frequency of 88 89 prescribed under burning in mixed pine stands. In both years of this study, plots burned every 2 years showed significantly higher activity for beetles than unburned plots. Likewise, higher levels of activity on plots burned every 6 years than on control plots were seen during 1994, but not during the following year. Thus it is possible to conclude that Carabid activity does not decrease with frequency of burning. If anything, these data suggest that activity diminishes with time since disturbance. The Carabid fauna of burned plots was significantly more diverse than unburned plots in a widely spaced, mixed pine stand. The ranking of diversity between the two burn treatments varied according to which element of diversity each model weighs. T-tests of the Shannon Diversity Index revealed significant differences between burned and unburned sites, but not between 2- and 6-year burn plots. Nevertheless, it is possible to conclude that Carabid species diversity does not decrease with increasing fire frequency in this pine ecosystem. The evidence neither supports the hypothesis that Carabid diversity is inversely related to time since disturbance, nor rejects it. Diversity estimates were very similar for plots burned at 2- and 6-year intervals. The data suggest that biennially burned plots may have greater abundance and species richness, and in some years lower species dominance. Despite having the lowest activity in 1995, diversity was higher on 6-year burn plots than 2-year plots according to the Berger-Parker, Shannon diversity and evenness, and the log series ot. This relationship probably reflects greater evenness of species abundance (lower 90 species dominance) on the 6-year burn plots in 1995, although the small sample size may have imparted some bias. This problem may be solved only with several more years of sample data, more observations per sample date, or longer trapping periods. Attempts to explain variation in beetle activity using vegetation characteristics produced marginal results and indicated complex relationships. Total Carabid activity was negatively related to seedling and sapling density, and unrelated to ground flora coverage, albeit with very low probability estimates. Analysis of count data for four common species was not helpful. Activity of P. stygicus was positively related to seedling and sapling density, and woody ground flora coverage, although most of these beetles were trapped on 2- year burn plots, where herbaceous coverage was highest, and other vegetation variables lowest. The resulting estimates and predicted relationships have very little predictive weight; probably due to small sample sizes (N < 150 for most species). It is also likely that the crudely 3-dimensional measurements used to describe vegetation characteristics, stem density and percent ground coverage by height and dbh class, inadequately characterized habitat elements from an insect’s point of view. Some researchers suggest that habitat suitability for insects may be more closely related to vegetation patchiness, floristic heterogeneity, and plant architecture (Southwood et al. 1979, Schowalter 1986, Lawton 1983). Regardless, my results do not clarify how beetle activity is related to vegetation coverage. Unfortunately, this portion of the study was to provide the pivotal evidence for rejection of the global null hypothesis, but this 91 cannot be done. It is clear that vegetation characteristics play an important part in determining beetle activity, but generalizations regarding species-specific activity would be fruitless given the strength of the equations that I have presented. Many studies have shown detrimental effects on insect fauna from wildfire and prescribed burning. Yet the results of studies involving wildfire may not be accurately compared to the lower-intensity disturbance created by controlled underburning. Earlier prescribed burning experiments cited herein may only reflect the magnitude of disturbance relative to site quality; most of the experiments were conducted in southern US. temperate forests on extremely sandy sites. My analysis suggests that controlled low intensity fire, when used for underburning in pine ecosystems on higher—quality sites, may have benign or even positive effects on Carabids and perhaps arthropod communities. LIST OF REFERENCES LIST OF REFERENCES Adis, J. 1979. Problems of interpreting arthropod sampling with pitfall traps. Zoologischer Anzeiger 202: 1 77-1 84. Ahlgren, |.F. 1974. The effect of fire on soil organisms. In T.T. Kozlowski and CE. Ahlgren, Eds., Fire and Ecosystems, Academic, New York, pp. 67-72. Alban, DH. 1977. Influence on soil properties of prescribed burning under mature red pine. USDA For. Serv. Res. Pap. NC-139. Alhgren, CE. 1976. Regeneration of red pine and white pine following wildfire and logging in northeastern Minnesota. J. For. 76:135-140. Alhgren, CE. 1979. Emergent seedlings on soil from burned and unburned red pine forest. Minn. For. Res. Notes 273. Anderson, N. 1964. Some relationships between grasshoppers and vegetation. Ann. Entomol. Soc. Am. 57:736-742. Anderson, R.C., Leahy, T. and SS. Dhillion. 1989. Numbers and biomass of selected insect groups on burned and unburned sand prairie. Am. Midl. Natur. 122:151-162. Baars, MA. 1979. Catches in pitfall traps in relation to mean densities of carabid beetles. Oecologia 41:25-46. Bergeron, Y., and J. Brisson. 1990. Fire regime in red pine stands at the northern limit of the species range. Ecology 71 :1352-1364. Bock, CE. and J.H. Bock. 1991. Response of grasshoppers (Orthoptera: Acrididae) to wildfire in a southeastern Arizona grassland. Am. Midl. Nat. 125:162-167. 92 93 Boswell, MT, and GP. Patil. 1971. Chance mechanisms generating the logarithmic series distribution used in the analysis of numbers of species and individuals. In Statistical Ecology, G.P. Patil, E.C. Pielou, and WE. Waters, eds. Pennsylvania State Univ. Press, University Park, PA. pp. 99-130. Bousquet, Y., and A. Larouchelle. 1993. Catalogue of the Gaedephaga (Coleoptera: Trachypachidae, Rhysodidae, Carabidae Including Cicindelini) of America North of Mexico. Mem. of the Entom. Soc. Can. No. 167. 397 pp. Brandmayr, T.Z. 1986. Spermophagous (Seed-eating) Ground beetles: First comparison of the diet and ecology of the Harpaline genera Harpalus and Ophonus (Coleoptera: Carabidae). In The Role of Ground Beetles in Ecological and Environmental Studies, eds. Intercept Ltd., Andover, Hampshire, UK. pp. 307-316. Briggs, J.B. 1961. A comparison of pitfall trapping and soil sampling in assessing populations of two species of Carabidae. East Malling Res. Sta. Ann. Report, 108-112. Brown, V.K. and T.R.E. Southwood. 1987. Secondary succession: patterns and strategies. Br. Ecol. Soc. Symp. No. 26. pp. 315—337. Buell, M.F., and J.E. Cantlon. 1953. Effects of prescribed burning on the ground cover in the New Jersey pine region. Ecology 34:520-528. Buffington, JD. 1967. Soil Arthropod populations in the New Jersey pine barrens as affected by fire. Entomol. Soc. Am. 60:530-535. Burgess, D.M., and IR. Methven. 1977. The historical interactions of fire, logging, and pine: a case study at Chalk River, Ontario. Info. Rep., Petawawa For. Exp. Stn., Canada. No. PS—X—66. 18 p. Collet, N.G., Neumann, F.G. and KC. Tolhurst. 1993. Effects of two short rotation prescribed fires in spring on surface-active arthropods and earthworms in dry schlerophyll eucalypt forest of west-central Victoria. Aust. For. 56;49-60. Dayton, PK. 1971. Competition, disturbance, and community organization: the provision and subsequent utilization of space in a rocky intertidal community. Ecological Monographs (41) pp. 351-389. Descender, K., Pollet, M. and R. Segers. 1984. Carabid beetle distribution along humidity-gradients in rivulet-associated grasslands (Coleoptera: Carabidae). Biol. Jb. Dodonaea. 52: 64-75. 94 Dickmann, D.I., 1993. Management of red pine for multiple use benefits using prescribed fire. North. J. Appl. For. 10:53-62. Dickmann, D.|., W.J. O’Niell, and N. Caveney. 1987. Wide-spaced red pine: A multiple use opportunity. North. J. Appl. For. 4:44—45. Einot, I. and KR. Gabriel. 1975. A study of the powers of several methods of multiple comparisons. Journal of the American Statistical Assn. 70: p. 351 Engstrom, PB, and DH. Mann. 1991. Fire ecology of red pine (Pinus resinosa) in northern Vermont, USA. Can. J. For. Res. 21:882-889. Evans, E. 1984. Fire as a natural disturbance to grasshopper assemblages of tallgrass prairie. Oikos 43:9-16. Evans, E. 1988. Community dynamics of prairie grasshoppers subjected to periodic fire: predictable trajectories or random walks in time? Oikos 55:283- 292. Evans, E.W. 1988. Grasshopper(lnsecta: Orthoptera: Acrididae) assemblages of tallgrass prairie: influences of fire frequency, topography, and vegetation. Can. J. Zool. 66: 1495-1501. Gieger, R. 1966. The climate near the ground. Harvard University Press, Cambridge, Mass. Gillon, D. 1972. The effects of bushfire on the principal pentatomid bugs (Hemiptera) of an Ivory Coast savanna. Tall Timbers Fire Ecol. Conf. Proc. 1 12377-417. Greenberg, CH. and MC. Thomas. 1995. Effects of forest management practices on terrestrial coleopteran assemblages in sand pine scrub. Fla. Entomol. 78:271-285. Greeslade, P. J. M. 1964. Pitfall trapping as a method for studying populations of Carabidae (Coleoptera). J. Anim. Ecol., 33:301-310. Hansen, JD. 1986. Comparison of insects from burned and unburned areas after a range fire. Great Basin Naturalist. 46: 712-727. Harris, D.L. and W.H. Whitcomb 1971. Habitat relationship and seasonal abundance of four species of Evan‘hrus (Coleoptera: Carabidae). Coleopt. Bull. 25:67-72. 95 Harris, D.L. and W.H. Whitcomb 1974. Effects of fire on populations of certain species of ground beetles (Coleoptera: Carabidae). Florida Entomologist 57197-103. Hawkins, BA. and EA. Cross. 1982. Patterns of refaunation of reclaimed strip mine spoils by nonterricolous arthropods. Environ. Entomol. 11:762-775. Hayter, A.J. 1984. A proof of the conjecture that the Tukey-Kramer method is conservative. Annals of Statistics, 12: p 61-75. Heinselman, ML. 1981. Fire and succession in the conifer forests of northern North America. In DC. West, H.H. Shugart, and DB. Botkin, eds. Forest succession: concepts and application. Springer-Verlag, New York, NY, USA. pp 347-405. Hejkal, J. 1985. The development of carabid fauna (Coleoptera: Carabidae) on spoil banks under conditions of primary succession. Acta Entomol. Bohemoslov. 82:321-346. Hengeveld, R. 1980. Qualitative and quantitative aspects of food of ground beetles (Coleoptera: Carabidae): a review. Neth. J. Zool. 30: 555-563. Henning, SJ, and DJ. Dickmann. 1996. Vegetative responses to prescribed burning in a mature red pine stand. North. J. Appl. For. 132140-146. Heyward, F. and AN. Tissot. 1936. Some changes in soil fauna associated with forest fires in the long leaf pine region. Ecol. 17:659-666. Hodgkins, E.J. 1958. Effects of fire on undergrowth vegetation in upland southern pine forests. Ecology 39:36-46. Holliday, NJ. 1991. Species responses of carabid beetles (Coleoptera: Carabidae) during post-fire regeneration of boreal forest. Can. Entomol. 123:1369—1389. Holliday, NJ. 1992. The carabid fauna (Coleoptera: Carabidae) during postfire regeneration of boreal forest: properties and dynamics of species assemblages. Can. J. Zool. 70:440-452. Holmes, P.R., Boyce, DC. and BK. Reed. 1993. The ground beetle (Coleoptera: Carabidae) fauna of Welsh peatland biotypes: factors influencing the distribution of ground beetles and conservation implications. Biol. Cons. 63: 153-161. 96 Hurst, GA. 1971. The effects of controlled burning on arthropod density and biomass in relation to bobwhite quail brood habitat on a right-of-way. Tall Timbers Conf. Ecol. Anim. Contr. Habitat Manage. Proc. 21173-183. Hutcheson, K. 1970. A test for comparing diversities based on the Shannon formula. J. Theor. Biol. 29:151-154. Jennings, D.T., Housewart, MW. and GA. Dunn. 1986. Carabid beetles (Coleoptera: Carabidae) associated with strip clearcut and dense spruce-fir forests of Maine. Coleopterists Bulletin 40: 251 -263. Johnson, NE, and RS. Cameron. 1969. Phytophagous ground beetles Ann. Entomol. Soc. Am. 62:909-914. Johnson, SR. 1995. Spider communities in the canopies of annually burned and long-term unburned Spartina pectinata wetlands. Environ. Entomol. 24: 832-834. Joosse, E. N. G. and J. M. Kapteijn, 1968. Activity stimulating phenomena caused by field disturbance in the use of pitfall traps. Oecologia, 12385-392. Jorgensen, JR. and CS. Hodges. 1970. Microbial characteristics of a forest soil after twenty years of prescribed burning. Mycologia. 62: 721-726. Kirk, V.M. 1972. Seed-caching by larvae of two ground beetles, Harpalus pennsylvanicus and H. erraticus. Ann. Entomol. Soc. Amer. 65:1426-1428. Kucera, C., and M. Koelling. 1964. The influence of fire on composition of central Missouri prairie. Am. Midl. Nat. 72:143—147. Lawton, J.H. 1983. Plant architecture and the diversity of phytophagous insects. Ann. Rev. Entomol. 28: 23-39. Lawton, J.H. and D. Schroder. 1977. Effects of plant type, size of geographical range, and taxonomic isolation on number of insect species associated with British plants. Nature. (265) pp. 137-140. Lemon, PC 1949. Successional responses of herbs in the longleaf—slash pine forest after fire. Ecol. 30:135—145. Lenski, RE. 1982. The impact of forest cutting on the diversity of ground beetles (Coleoptera: Carabidae) in the southern Appalachians. Ecological Entomology 71385-390. 97 Liebherr, J. and Mahar, J. 1979. The carabid fauna of the upland oak forest in Michigan: survey and analysis. The Coleopterists Bulletin (33) 183—197. Lindroth, C.H. 1961-1969. The Ground-Beetles (Carabidae, excl. Cicindelinae) of Canada and Alaska, Parts 1-6. Opuscula Entomol. XX, XXIV, XXIX, XXXIII, XXXIV, XXXV. (suppl.) Loreau, M. 1983. Trophic role of carabid beetles in a forest. In Proceedings of the VIII lntl. Colloquium of Soil Zoology. Louvain—Ia-Neuve, Belgium, August 30-September 2, 1982. Ph. Lebrun, H.M. Andre’, A. De Medts, C. Gregoire- Wibo, and G Wauthy, eds. lmprimeur Dieu-Brichart, Ottignes-Louvain-la- Neuve. pp. 281-286. Loreau, M. 1986. Niche differentiation and community organization in forest carabid beetles. In Carabid beetles: their adaptations and dynamics. Edited by P.J. den Boer, M.L. Luff, D. Mossakowski, and F. Weber. Gustav Fischer, Stuttgart. pp. 465—487. Luff, ML. 1975. Some factors influencing efficiency of pitfall traps. Oecologia 19:345-347. Lunt, HA. 1950. Liming and twenty years of litter raking and burning under red (and white) pine. Soil Sci. Soc. Am. Proc. 381-390. Lussenhop, J. 1976. Soil arthropod responses to prairie burning. Ecology. 57: 88-98. Lyon, L.J., Crawford, H.S., Cuzuhai, E., Fredriksen, R.L., Harlow, F.R., Metz, L.J., and HA. Pearson. 1978. Effects of fire on fauna: a state-of—knowledge review. us. Dep. Agric. For. Serv. Gen. Tech. Rep. WO—6. Mader, H.J. 1986. The succession of carabid beetles in a brown coal mining area and the influence of afforestation. In Carabid beetles: their adaptations and dynamics. Edited by P.J. den Boer, M.L. Luff, D. Mossakowski, and F. Weber. Gustav Fischer, Stuttgart. pp. 497-508. Magurran, A. E. 1988. Ecological Diversity and Its Measurement. Princeton University Press, Princeton, N. J. Maissarow, D.K. 1941. The role of fire in the perpetuation of virgin forests of northern Wisconsin. J. For. 39:201-207. Manley, G.V. 1971. A seed-caching carabid (Coleoptera). Ann. Entomol. Soc. Amer. 64:1474-1475. 98 May, RM. 1975. Patterns of species abundance and diversity. In Ecology and Evolution of Communities, M.L. Cody and J.M. Diamond, eds. Harvard Univ. Press, Cambridge, MA. pp. 81-120. McCoy, ED. 1987. The ground-dwelling beetles of periodically burned plots of sandhill. Fla. Entomol. 70: 31-39. McCullough, D.G., and HM. Kulman. 1991. Differences in foliage quality of young jack pine (Pinus banksiana Lamb.) on burned and unburned sites: effects on jack pine budworm (Choristoneura pinus pinus Freeman). Oecologia 87:135-145. Methven, IR, and W.G. Murray. 1974. - Using fire to eliminate understory balsam fir in pine management. For. Chron. 50:77-79. Metz, L.J. and D. Dindal. 1975. Collembola populations and prescribed burning. Environmental Entomology. 41583-587. Metz, L.J. and M.H. Farrier. 1973. Prescribed burning and populations of soil mesofauna. Environ. Entomol. 2:433—440. Miller, W.E. 1978. Use of prescribed burning in seed production areas to control red pine cone beetle. Environ. Entomol. 7:698—702. Mitchell, R.G. and RE. Martin. 1980. Fire and insects in pine culture of the Pacific Northwest. Proc. 6th Conf. Fire. For. Meteor., Soc. Am. For., pp. 182- 190. Moya-Raygoza, G. 1995. Fire effects on insects associated with the gamagrass Tripsacum dacty/oides in Mexico. Ann. Entomol. Soc. Am. 88:434-440. Murdoch, W.W., Evans, PC, and CH. Peterson. 1972. Diversity and pattern in plants and insects. Ecology. 53:819—829. Neumann, U. 1971. Die Ausbreitungsfahigkeit von Arabiden in den forstlichen Rekultivierungen des rheinsischen Braunkohlenreviers. Misc. Pap. Landbouwhogesch. Wageningen, 8289-102. Neve, G. 1994. Influence of temperature and humidity on the activity of three Carabus species. In K. Desender et al. (eds), Carabid Beetles: Ecology and Evolution. Kluwer Academic Publishers, Netherlands. pp. 189-192. Niemala, J., Halme, E., Pajunen, T. and P. Punttila. 1990. Diversity variation in carabid beetle assemblages in the southern Finnish taiga. Pedobiologia 34:1- 10. 99 Niemala, J., Langor, D. and JR. Spence. 1993. Effects of clear-cut harvesting on boreal ground beetle assemblages (Coleoptera: Carabidae) in western Canada. Conservation Biology 7:551-561. Niemela, J., Halme, E. and Y. Haila. 1990b. Balanced sampling effort in pitfall trapping of carabid beetles. Entomologica Fennica 12233-238. Niemela, J., Halme, E., Pajunen, T. and Y. Haila. 1986. Sampling spiders and carabid beetles with pitfall traps: the effect of increased sampling effort. Annales Entomologici Fennici 522109-111. Niering, W.A., R.H. Goodwin, and S. Taylor. 1970. Prescribed burning in southern New England: Introduction to long-range studies. Proc. Tall Timbers Fire Ecol. Conf. 10:267-286. Pearse, AS. 1943. Effects of burning over and raking off litter on certain soil animals in the Duke Forest. Am. Midl. Nat. 29:405—424. Pyne, SJ. 1982. Fire in America. Princeton Univ. Press, Princeton, NJ. 654 p. Reeves, R.M., Dunn, GA. and D.T. Jennings. 1983. Carabid beetles (Coleoptera: Carabidae) associated with the spruce budworm, Choristoneura fumiferana (Lepidoptera: Tortricidae). Can. Entomol. 115:453-472. Reifsnyder, W.E., L.P. Herrington, and K.W. Spalt. 1967. Thermophysical properties of bark of shortleaf, longleaf, and red pine. Yale Univ. Sch. For. Bull. No. 70. 41 p. Rice, LA. 1932. The effect of fire on the prairie animal communities. Ecol. 13:392-401. Richardson, R.J., and NJ. Holliday. 1982. Occurrence of carabid beetles (Coleoptera: Carabidae) in a boreal forest damaged by fire. Can. Ent. 114:509-514. Rickard, W.H. 1969. Ground dwelling beetles in burned and unburned vegetation. J. Range. Manage. 23:293-294. Riechert, S. and W. Reeder. 1972. Effects of fire on spider distribution in southwestern Wisconsin prairies, p. 73-90. In J.Zimmerman (ed.). Proc. 2nd Midwest Prairie Conf., Madison, Wisconsin. Rouse, C. 1988. Fire effects in northeastern forests: Red pine. USDA For. Serv. Gen. Tech. Rep. NC-129. 100 Schowalter, T.D., Hargrove, W.W., and DA. Crossley. 1986. Herbivory in forested ecosystems. Ann. Rev. Entomol. 31:177-196. Simmons, GA. 1977. Preliminary test of prescribed burning for control of maple leaf cutter (Lepidoptera: lncurvariidae). Great Lakes Entomol. 10:209-210. Snider, RM. and R.J. Snider. 1986. Evaluation of pit-trap transects with varied trap spacing in a northern Michigan forest. Great Lakes Entomol. 19:51 -61. Sousa, WP. 1979. Disturbance in marine intertidal boulder fields: the nonequilibrium maintenance of species diversity. Ecology (60) pp. 1225-1239. Southwood, T.R.E. 1978. Ecological Methods, Chapman and Hall, London. Southwood, T.R.E., Brown, V.K. and PM. Reader. 1979. The relationships of plant and insect diversities in succession. Biol. J. Linn. Soc. 12:327-348. Stokes, ME, Davis, CS, and G.G. Koch. 1995. Categorical Data Analysis Using the SAS System. Cary, NC: The SAS Institute Inc., 499 pp. Szyszko, J. 1987. How can the fauna of Carabidae be protected in managed pine forest? Acta Phytopath. Entom. Hung. Vol. 22(1-4) pp. 293-303. Tester, JR. and W.H. Marshall. 1961. A study of certain plant and animal interrelationships on a prairie in northwestern Minnesota. Univ. Minn., Mus. Natur. Hist, Occas. Pap. No. 8. Thiele, H.U. 1977. Carabid beetles in their environment. Springer-Verlag, Berlin. Trabaud, L. 1987. The role of fire in ecological systems. SPB Acad. Publ., The Hague, The Netherlands. Van Wagner, CE. 1970. Fire and red pine. Proc. Tall Timbers Fire Ecol. Conf. 10:21 1-219. Viereck, LA, and CT. Dyrness. 1979. Ecological effects of the Wickersham Dome fire near Fairbanks, Alaska. USDA For. Serv. Gen. Tech. Rep. PNW- 90. p. 53-55. Whelan, R.J., Langedyk, W. and AS. Pashby. 1980. The effects of wildfire on arthropod populations in jarrah-Banksia woodland. Western Australian Naturalist. 14: 214-220. 101 White, AS. 1983. The effects of thirteen years of annual prescribed burning on a Quercus ellipsoidalis community in Minnesota. Ecology 64:1081-1085. Williams, CB. 1964. Patterns in the Balance of Nature and Related Problems in Quantitative Ecology. Academic Press, London. Winter, K. von. 1980. Sukzession von Arthropoden in verbrannten Kiefernforsten. Ill. Laufkafer (Carabidae). Forstwiss. Centralbl. 99:356- 365.Greenslade, P.J.M. 1964. The distribution, dispersal, and size of a population of Nebria brevico/lis (F.) with comparative studies on three other Carabidae. J. Animal Ecology "IllllllllllllllllllllllllS