Your unpublished thesis, submitted for a degree at Williams College and administered by the Williams College Libraries, will be made available for research use. You may, through this form, provide instructions regarding copyright, access, dissemination and reproduction of your thesis. &The faculty advisor to the student writing the thesis wishes to claim joint authorship in this work. In each section, please check the ONE statement that reflects your wishes. I. PUBLICATIONAND QUOTATION: LITERARYPROPERTYRIGHTS A student author automatically owns the copyright to hisiher work, whether or not a copyright symbol and date are placed on the piece. The duration of U.S. copyright on a manuscript--and Williams theses are ' considered manuscripts--is the life of the author plus 70 years. -b'we do not choose to retain literary property rights to the thesis, and I wish to assign them immediately to Williams College. Selecting this option will assign copyright to the College-This in no way precludes a student author Erom later publishing hisher work; the student would, however, need to contncl the Archives for a pcrrnisGon li)rn~.The ikchives would be free in this case to also grant perru~ssion to another researcher to ptrhlish srniill sections from [he thesis. Rmly would there be any reaboa for the i'lrchivcs io grant permission lo another party to ptlhlish thc ti1es:s iniLs endrely; 11"such a situation arobe, the Archives wodd be in touch with thc aulhor to Ict thcrn know that such d request had becn ma&. -Uwe wish to retain literary property rights to the thesis for a period of three years, at which time the literary property rights shall be assigned to Williams College. Selecting this optjon gives the author a few years to l-nake e;uclusivc use of the t11e:;is in up-cttm~~lg pmjzcts: articles, later research, ctc, _2( Iiwe wish to retain literary property rights to the thesis for a period of ID years, or until my death, whichever is the later, at which time the literary property rights shall be assigned to Williams College. Selecting lhls option allctws tllo autl~orgat flexibility in extending or shortening tllc time of bis/her automatic copyright period. Some students areinrerestcd in using their thesis in graduatc school work. fr, this cssc, it would zmke: sense for them to enta. a number such as '10 years' in thc blanlc, and linc out the words 'or until my death. whichever is the later.' In any cvcnt, it is easier for the Archives to administer copyright on a r~lanuscripltf tflc period ends wid1 the individual's death-i)ur staff won't have to search for csbtc executors in tlris case--tmt this is entirely up to ench student. LI. ACCESS The Williams College Libraries are investigating the posting of theses online, as well as their retention in hardcopy. _y_ Williams College is granted permission to maintain and provide access to my thesis in hardcopy and via the Web both on and off campus. Selecting this option allows rese~irchers sound the world to access the digital ve~sion of your work. -Williams College is granted permission to maintain and provide access to my thesis in hardcopy and via the Web for on-campus use only. Selecting this option allows access to the digital version of your work from the on-campus network only. -The thesis is to be maintained and made available in hardcopy form only. Selecting this option allows access to your work only from the hardcopy you submit. Such access pertains to the entirety of your work. inciuding any media that it conlprises or includeh. 111.COPYINGAND DISSEMINATION Because theses are listed on FRANCIS, the Libraries receive numerous requests every year for copies of works. Iflwhen a hardcopy thesis is duplicated for a researcher, a copy of the release form always accompanies the copy. Ar,y digital version of your thesis will include the release form. -X Copies of the thesis may be provided to any res&her. Selecting this option allot~s any researcher to request a copy from the Williams College Libraries, or to make one from an electronic version. -Copying of the thesis is restricted for -years, at which time copies may be provided to any researcher. This option allows the autfior to set a time limit on copying restriclions. During this period, an electroilic version of the thesis will be protzcted against duplication. -Copying of the thesis or portions thereof, except as needed to maintain an adequate number of research copies available in the Williams College Libraries, is expressly prohibited. The electronic version of the thesis will be protected against duplication. Selecting this option allows no reproductions to be made for researcllers. The electronic version of the thesis will be protected against duplication. This option does not dis-allow researchers horn readingfviewing the work in either hardcopy or digital form Date accepted Land-use history and the invertebrate decomposer communities of eastern deciduous forests by William C. Wetzel Henry W. Art, Advisor A thesis submitted in partial fulfillment of the requirements for the Degree of Bachelor of Arts with Honors in Biology WILLIAMS COLLEGE Williamstown, Massachusetts May 21,2006 Table of Contents Abstract....................................................................... 5 Introduction..................................................................7 The decomposer food web ........................................9 Literature Review ............................................................11 Organic inputs and forest succession ............................11 Chronosequences................................................... 14 Immediate land-use effects ....................................... 15 Nutrients and succession .......................................... 17 Habitat heterogeneity ..............................................18 Summary........................................................... 18 Settings........................................................................ 20 Location and climate ...............................................20 Geology and soils ...................................................21 Vegetation...........................................................22 Land-use history ....................................................23 Methods...................................................................... 26 Study sites ........................................................... 26 Invertebrate sampling ..............................................28 Invertebrate extraction .............................................29 Invertebrate identification ......................................... 31 Soil chemistry and texture ........................................ 32 Statistical methods ................................................. 33 Results and discussion ......................................................35 Invertebrates......................................................... 35 Forest type ...........................................................38 Elevational gradients ...............................................43 Land-use and invertebrate communities ........................46 The top predator ....................................................50 Environmental influences .........................................59 Variance.............................................................64 Conclusions..................................................................67 Acknowledgements......................................................... 69 References.................................................................... 71 Appendix..................................................................... 77 This thesis compared the forest-floor invertebrate communities in secondary (post- agricultural) and primary (never completely cleared) eastern deciduous woodlots in northwestern Massachusetts. Invertebrates were extracted with an enhanced Tullgren- Berlese funnel from litter and soil samples collected at 77 forest plots. Individuals greater than 2mm in any dimension were identified to order. Spiders (Araneida) were significantly more abundant in primary woodlots than in secondary woodlots, but overall invertebrate density did not differ significantly between forest types. Spiders, as the top predator in the invertebrate food web, can serve as indicators of overall community structure. Their dearth in secondary woodlots suggests that the forest-floor communities are less complex on post-agricultural land, with less energy reaching the upper trophic levels. Quality of leaf litter in primary woodlots may explain this difference. Spider density increases with leaf litter depth in primary woodlots, but has no relationship with litter depth in secondary woodlots. Sugar maple (Acer saccharum) is more abundant in primary woodlots, whereas red maple (Acer vubrum) and paper birch (Betulapapyrifera) favor secondary woodlots. Soils are more alkaline in primary woodlots and this is linked to forest-canopy composition. Despite differences in spider densities, significant overall differences were not found between the invertebrate communities in primary and secondary woodlots. The order level of taxonomic resolution may not have been fine enough to show the community differences beyond that of spider density. HNTRBDUCTILBN Despite extensive study on aboveground land-use ecology (Braun 1950, Art and Dethier 1986, Singleton et al. 2001, McBride 2006), little is known about how human land-use affects the belowground decomposer communities in the long term (Wardle 2002). A lack of understanding of the belowground system is critical given that the decomposer food web is an essential component of terrestrial ecosystems (Wall and Moore 1999, Ad1 2003, Bardgett 2005). This study compares the forest floor macroinvertebrate communities of primary (never completely cleared) deciduous, primary hemlock, and secondary (post-agricultural) forest stands across a range of floristic and abiotic variables. This thesis hypothesized that hemiedaphic (litter and upper soil layer) invertebrate communities would differ in primary and secondary woodlots and that environmental variables like elevation would also play an important role in structuring these communities. The forest floor community is a vital component of forest ecosystems. Hairston et al. (1960) made a simple observation about energy flow in terrestrial ecosystems that encouraged ecologists to look more closely at the forest floor. They noticed aboveground herbivores consume a small fraction of the available green matter, yet only a negligible fi-action of that unconsumed organic matter is fossilized. This means that more than half of the energy in a deciduous forest passes through the decomposer food web (Weary and Merriam 1978). The importance of the forest floor is even greater than the huge amounts of energy it consumes. Wardle (2002, p. 1) explains that the producer and decomposer subsystems are "obligately dependent upon one another." The decomposer subsystem converts complex organic materials into inorganic compounds. This process releases and cycles nutrients that otherwise would be locked in downed organic debris (Paquin and Coderre 1997, Wardle 2002). Using mesh litterbags to exclude fauna, Gonzales and Seastedt (2001) showed that hemiedaphic invertebrates directly enhance soil nutrient concentrations and consequently forest productivity. Without saprophytes, dead organic matter would accumulate ad nauseam, stifling terrestrial ecosystems (Milcu et al. 2006). To quantify the importaiice of the hemiedaphic arthropod community, Weary and Merriam (1 978) applied insecticide to leaf litter in a red maple woodlot in southeastern Canada. The reduction in the arthropod community resulted in depressed rates of decomposition and an accumulation of organic detritus. Despite their fundamental role in all terrestrial ecosystems, the hemiedaphic invertebrates are not well understood. In a review on sampling techniques, AndrC et al. (2002) conclude that less than 10% of soil microarthropod species have been described. Even with such a small fraction of the fauna classified, it is estimated that the soil and litter system is second in diversity to only coral reefs (Strange 1997). A clear majority of terrestrial invertebrate species spend all or a significant portion of their lives belowground (Wardle 2002). In well developed temperate forest soils, mites and springtails alone may reach populations of millions m-2 (Anderson 1975, Blair et al. 1994). The decomposer food web The forest-floor decomposer food-web is complex and contains multiple, often indistinct trophic levels (Ponsard et al. 2000). Initially, energy from primary production reaches the forest floor as organic detritus: leaf litter, woody debris, and animal feces or carcasses. Fungi and bacteria are the main primary-saprotrophs and are responsible for initiating decomposition (Bardgett 2005). Through extracellular digestion, they make the energy and nutrients stored in dead organic matter accessible to the soil environment (Ad1 2003). The secondary saprotrophs consist of protozoa, potworms (Enchytraeids), nematodes (Nematoda), and the microarthropods: mites (Acari) and springtails (Collembola). The species in this trophic level are categorized as bactivorous, fungivorous, or detritivorous, but in truth these groups are blurred as many species are opportunists. Secondary saprotrophs often consume detritus along with whatever bacteria or fungi the organic matter supports (Ad1 2003) Macroarthropods, isopods, annelids, and gastropods occupy the next trophic level. These groups overlap in food preferences with the secondary saprotrophs. Some consume detritus and, like the secondary saprotrophs, eat whatever is growing on it (Ad1 2003). Other species are more directly predaceous, however even predator-herbivore categories are indistinct. Kounda-Kiki et al. (2004) describe an oribatid mite species that turned from fungivory to predation in a nitrogen poor environment. The decomposition process depends on the interactions between all of these trophic levels (Wall and Moore 1999). The larger invertebrates are responsible for the mechanical breakdown of detritus. Annelids and detritivorous arthropods consume large pieces of organic material and excrete finely shredded fecal pellets that are high in surface area and moisture. Primary saprotrophs are found at significantly higher densities in fecal pellets than in undigested organic matter (Ad1 2003, Bardgett 2005). As evidence of the magnitude of this process, the earthworms inhabiting temperate ecosystems excrete 20-40 ton of feces per hectare (Edwards and Bohlen 1996). Predation is a major trophic interaction on forest floors and has potential to exert top- down control on rates of decomposition (Scheu et al. 2003). Predators in the soil food web include nematodes, mites, spiders (Araneida), pseudoscorpions (Pseudoscorpionida), beetle larvae (Coleoptera), centipedes (Chilopoda), and ants (Formicidae). Santos et al. (1 98 1) found predatory mites suppress bactivorous nematode populations and indirectly increase microbe populations and rates of decomposition. Conversely, Mikola and Setala (1998) showed that in a microcosm experiment microbial populations were independent of predator populations. The secondary saprotrophs, however, were significantly regulated by predation. Though the processes are not yet well understood, it is clear that predators are important members of the decomposer food web. LPTlfEMTUm mWEW Most researchers interested in land-use and hemiedaphic invertebrate communities have considered only the short-term effects of silvicultural or agricultural practices (Seastedt and Crossley 1981, Greenberg and McGrane 1996, Niemela 1997, Neave and Fox 1998, Yeates et al. 2000, Barros et al. 2002, Raty and Huhta 2004, Bird et al. 2004). The few papers that have addressed long-term land-use effects tend to focus on one taxonomic group, often springtails (Addison et al. 2003, Hasegawa et al. in press). Although these studies do not resolve the effects of land-use on invertebrate communities, they do begin to characterize some of the forces that are important community-assembly forces. Organic inputs and forest succession Quality and abundance of organic matter has been found to be a significant factor in structuring decomposer invertebrate communities in many studies (e.g. McBrayer et al. 1977, Burghouts et al. 1992, Blair et al. 1994, Ponsard et al. 2000, Scheu et al. 2003, Jabin et al. 2004, Kounda-Kih et al. 2004). These studies often link forest succession with the quantity and decomposability of organic inputs. Late successional and old- growth forests tend to have more available organic matter (McBrayer et al. 1977, Blair et al. 1994, Paquin and Coderre 1997). McBrayer et al. (1977) were among the first to systematically compare soil web communities in different forest types. They show that both invertebrate populations and litter quantity in mature temperate Douglas fir and deciduous forests greatly exceed those of a Douglas fir plantation and secondary deciduous forest, respectively. Successional stages, including plantations, attain maximum productivities within a relatively short period but standing crop increases well beyond the attainment of maximum productivity (Whittaker 1970). This implies that decomposition is less than net production during the development of a forest and if there is less to decompose, one would expect fewer organisms to decompose it. McBrayer et al. (1977, p.91) The importance of organic matter is apparent, but studies comparing it with forest succession and invertebrate communities have had mixed results. Downed organic debris represents two resources for the decomposer community. First, detritus is the main source of food and energy. Second, litter and woody debris provide habitat space. Because measurements of litter quantity take both of these attributes into account, it is difficult to distinguish their relative importance. Scheu et al. (2003) determine that litter food quality is more important to the decomposer community than overall litter quantity. They compared European beech (Fagus sylvatica) and Norway spruce (Picea abies) forests of different ages. Far more organic matter was present in the spruce forests, but European beech litter was more palatable to invertebrates than spruce needles. Consequently, invertebrate abundance and a-diversity were greater in the beech forest (Scheu et al. 2003). This also explains the greater accumulation of litter in the spruce forests. Scheu et al.'s (2003) conclusion that beech leaves are better food for decomposers than spruce needles is based on the assumption that the two substrates provide equivalent habitat quality. It is possible that the limiting factor for decomposers in needle substrate is moisture. A needle litter layer is more porous than a leaf litter layer and would dry out more easily than a thick layer of leaves. To truly understand the structuring forces in these complex hemiedaphic communities, studies need to measure an array of chemical and physical litter parameters. Blair et al. (1994) compared four forest stands (a mid-successional mixed deciduous stand, a young red pine plantation (Pinus vesinosa), a late successional beech stand, and a ca. 300 year old hemlock, Tsuga canadensis, stand) in east central New York and found that the hemlock stand had the highest standing stock of organic matter and the highest microarthropod densities. Unfortunately, the low replication and forest differences aside from stand age prevent robust successional interpretations. In addition, it is unlikely that the other three stands will be replaced by hemlocks as they age and therefore there is no successional link between these sites. Nonetheless their results contradict those of Scheu et al. (2003). Springtail numbers were 4-8 times greater in the needle litter of the hemlock stand than in all other forest types. Blair et al. (1994) positively correlate springtail and mite abundances with standing stocks of organic matter, regardless of litter recalcitrance. They state "accumulation of organic matter may be the most important determinant of local microarthropod abundances in these forests." Blair et al. (1 994) also examined moisture and found it to be positively associated with invertebrate densities. Jabin et al. (2004) looked at coarse woody debris instead of litter. They found strong positive associations between the macroarthropods and coarse woody debris. In a German oak-beech forest, macroarthropod populations and diversities were twice as great sites high in coarse woody debris. Even within stands, macroarthpod communities were larger and more complex near woody debris. They claim that in Central Europe, undisturbed forests contain 17-200 times more coarse woody debris by volume than managed forests. Based on the limited current knowledge, Wardle (2002, p.99) hypothesizes that large invertebrates (probably >5mm) will play "an increasing role" in the decomposer food web as the "maximal biomass" stage is reached late in succession. He suggests that fungi will become the dominant primary saprotrophs. Because fungi are larger and easier to graze, they may support a much more complex food web than prokaryotes do earlier in succession. Chronosequences Paquin and Coderre (1997) tried to measure the long-term effects of succession on the decomposer community. They sampled the macroarthropods of three boreal forest stands which were last burned in 1760, 1847, and 1944. This chronosequence documents the local deciduous to coniferous successional gradient. Contrary to McBrayer et al. (1977), Paquin and Coderre (1997) find a decrease in abundance and diversity through succession. Regrettably, their study suffers from pseudoreplication. They defend their study by claiming that abiotic differences between their 3 sites are negligible. A glance at the map they present reveals that the 1944 and 1847 sites are situated on the south shore of a lake approximately Ikm apart, while the 1760 forest stand is 9km north, on the north shore of the same lake. There may or may not be significant spatial autoconelation between the two close sites, but there are definitely environmental differences between the north and south shores of a large lake. Without replication of forest stands, it is impossible to know if faunal differences are due to stand age or confounding environmental variables. Forests that appear identical to even the most observant ecologist may look very different through compound eyes. Two studies (Addison et al. 2003 and Hasegawa et al. in press) measured springtail densities along chronosequences of temperate forests. Addison et al. (2003) sampled a chronosequence of four stands replicated in three Douglas-fir forests. They found that 80-102 years post-disturbance, secondary woodlots supported springtail populations that were significantly lower than the ones in old-growth forests. Addison et al. (2003) failed to measure important environmental variables such as litter abundance, soil moisture, and soil nutrient concentrations and therefore their study is missing much of the picture. Hasegawa et al. (in press) sampled eight deciduous stands from 1 to 128 years old with less significant results. Springtail numbers were significantly lower at the I year old stand once out of the three times during seven months they sampled the chronosequence. From this one instance of statistical significance, the authors concluded that springtail populations in these deciduous forests must return to pre-cutting levels within 4 years. Other than this one significance, there was no noticeable trend in springtail populations. Clearly, more research is required before conclusions can be drawn. Immediate land-use effects Burghouts et al. (1992) compared a primary and a selectively logged dipterocarp forest in Malaysia. The results need to be interpreted carefully because there was no replication, but the primary woodlot supported a greater abundance of litter invertebrates. This was correlated with leaf litter and fine root mass. On the other hand in the logged forest, invertebrate abundances correlated with litter humidity and woody debris, not with litter layer mass. This indicates that different forces are structuring the communities in the primary and logged forests. Unlike many other studies, Burghouts et al. (1992) looked at the entire community and thus revealed the complexities of the system. Although the overall community was more abundant in prima17 woodlots, beetles, millipedes (Diplopoda), and cockroaches (Blattodea) were more abundant in the selectively logged forest. In the short-term, agriculture and silviculture significantly alter ecosystems. Forest clear- cutting and conversion to agriculture result in a decrease in the amount and complexity of soil organic matter. In addition, physical disturbance to soils and enhanced erosion is significant (Bardgett 2005). In West Africa, Eggleton et al. (2002) measured termite (Isoptera) a-diversity along an anthropogenic disturbance gradient. There were fewer than half as many termite species at agricultural sites as at forest sites. Forestry, on the other hand, does not seem to affect macroinvertebrates to the same extent as agriculture. An agroforestry plantation of almond trees contained nearly as many termite species as a primary woodlot. Similarly, Bloemers et al. (1997) found that slash and burn supported fewer than half the number of tropical nematode species compared to a near primary or an old secondary woodlot. Nutrients and succession Deleporte and Tillier (1999) sampled the soil fauna in plots to which CaC03, N (ammonitrate), and NPKCa (ammonitrate, potassium sulfate, P205, and CaC03) had been applied in 1973 in a controlled randomized study. 22 years later, the nutrients amendments still had significant effects on the soil structure and invertebrate communities. The humus morphology of the CaCOj and NPKCa plots changed from a moder with distinct organic litter layers to a mull-moder humus type with less litter and poorly developed layers. Mull and moder structures typically support very different invertebrate communities (Schaefer and Schauermann 1990, Paquin and Coderre 1997). Deleporte and Tillier (1 999) found lower numbers of springtails and mites in N enriched plots. The rest of their invertebrate results are unclear and mostly lacked statistical significance. While his interpretations are mainly limited to the primary saprotrophs, Wardle (2002) interprets the forest floor successional pathway based on limiting nutrients. Contrary to other researchers, he explains that organic matter accumulates more rapidly earlier in forest regeneration, but microflora populations are limited by low N levels. Late in succession, bacterial fixation makes N abundant in the soils and consequently, the primary saprotroph populations grow until organic carbon becomes limiting. The result, more accessible carbon in late successional stages, is the same but the mechanism is very different. Future studies should measure soil N concentrations to better understand these processes. Habitat heterogeneity Old-growth Fennoscandian boreal forests support a high diversity of invertebrates. Niemela (1997) attributes this diversity to micro-habitats provided by abundant coarse woody debris and disturbance-gap dynamics. He claims that although local a-diversity may increase after a logging event, regional a-diversity is lost. This change in composition is because generalist species invade freshly cut forests, but old-growth specialists become locally extinct. McBride (2006) explains that agriculture reduces habitat heterogeneity. Plowing and grazing reduces microtopographic variations that are abundant in primary woodlots. When mature trees fall, their roots expose large pits and their trunks decay and form mounds. The result is the mounding and pitting pattern typical of primary woodlots (Saterson 1977, Reiss 2000, Simpson 2000). Summary Overall, results from studies on forest succession and decomposer communities are mixed. Niemela (1997) found that old-growth forests support species specialized to microhabitats not present in secondary woodlots. Eggleton et al. (2002), on the other hand, found that a plantation and a primary woodlot contained nearly the same species composition. McBrayer et al. (1977) conclude that richness and abundance in mature Douglas fir forests is greater than secondary woodlots, but Hasegawa et al. (in press) claim that springtail communities return to predisturbance states in four years. These studies mainly compare immature and mature forests, whereas this thesis compares forests with differing land-use histories. Simple succession chronosequence studies may be poor predictors of land-use effects, but they are all that is available in the literature. The changes that agriculture imposes on decomposition may not be independent of forest succession. Past studies show a variety of factors that are important in structuring forest-floor invertebrate communities. Organic matter quality and quantity appear to be of significance. Other studies hint that invertebrates have nutrient availability or moisture requirements. Microtopography creates habitat heterogeneity that encourages diversity. SETTINGS Location and climate Williamstown is situated in the northwest comer of the Berkshires and Massachusetts. Three mountain ranges converge within the town's 121krn2. The Taconic kdge (700- 850m) forms the dramatic western border, while the Greylock Massif (750-1063m) dominates the southeast. To the north, is the southern terminus of the Green Mountains. The topography is undulating in the valleys and steep in the hills. The elevation ranges from 200m on average in the valleys to 1063m at the summit of Mount Greylock, the roof of Massachusetts. The Hoosic River empties Williamstown's watersheds and meanders north through Vermont and New York into the Hudson River. The area has a humid continental climate. July and January are the warmest and coldest months, averaging -6°C and 20°C respectively. At high elevations, snow usually persists from December to March. Precipitation comes uniformly through the seasons and averages approximately lOOOrnm per year (Ai-t and Dethier 1986). Willianistown has a significant amount of forested area under state and town protection. The Taconic Range contains the Taconic Trail State Park and the Greylock Massif is in the Mount Greylock State reservation. The Williamstown Rural Lands Foundation protects land in the valleys and in the southern Green Mountains. Williams College owns the Hophns Memorial Forest, a 10krn2 research and educational forest which sits in the northwest comer of Massachusetts. Geology and Soils In the Taconic Orogeny during the Ordovician, volcanic island arcs collided with the North American Plate, mixing a variety of marine substrates together. The result is the varied geology of Williamstown. Phyllite and schist, high in quartz and feldspar, comprise the bedrock of the Taconic Range in western Williamstown. The Greylock Massif to the south and east has higher-grade schist, with less quartz and feldspar (Art and Dethier 1986). The Greylock Massif has carbonates in the form of dolostone and dolomitic phyllite, but the Taconic Range's carbonates are limestone. The carbonates are abundant at lower elevations, though carbonitic bands are found at a variety of elevations. The southern reaches of the Green Mountains in northeast Williamstown are predominantly quartzite. The valleys, on the other hand, contain weathering calcitic dolomite marble, rich in quartz (Zen 1983). The surface geology is a product of Pleistocene-Holocene glaciation cycles. Soils are mineralogically similar to the local bedrock and overlaying glacial till in most areas. Glacial till forms a 1-3m thick layer in most places. Till and soils contain slightly more quartz than the local bedrock, possibly due to weathering. According to Art and Dethier (1986), soils formed in the past 14,000 years, are 70-120cm thick and generally acidic. As much as 50% of soils over carbonate bedrock may be allocthonous fragments, but these transported materials consist of schist and quartz and are comparable to the dominant regional soil character (Art and Dethier 1986). Local soils are inceptisols and spodosols. Spodosols form fi-om coarse sandy, quartz enriched materials. They are characterized by leaching of nutrients from the surface into lower mineral layers. They are generally infertile, acidic, and associated with hemlock stands in this region, but also support deciduous species. Hemlocks and other conifers often encourage formation of spodosols through their recalcitrant, acidic litter (Lutz and Chandler 1966, Soil Survey Staff 1999). Compared to spodosols, inceptisols show less- defined horizons. They are typically less acidic and contain more loam and clay. Inceptisols have an ill-defined boundary between organic and mineral materials. This mixing is often associated with earth worm action. In this region inceptisols commonly support mixed decidous forests (Lutz and Chandler 1966, Soil Survey Staff 1999). Vegetation Vegetational succession in eastern deciduous forests is well studied (Braun 1950, Hibbs 1983, Reiss 2000, Simpson 2000, Singleton et al. 2001, McBride 2006). Foresters can predict that paper birch (Betula papyrgera) and other r-selected species will be the first trees to colonize abandoned agricultural land. Red maple (Acer rubrum) and red oak (Quercus rubra) after approximately 75 years will shade out the colonizing species. Finally, shade tolerant beech (Fagus grandifolia) and sugar maple (Acer saccharurn) will dominate old stands (Artand Dethier 1986). However, tree recolonization is not an indication that all components of a forest ecosystem have recovered. Many studies show that the forest-herb guild is significantly slower than the woody guild at recolonizing post agricultural fields (Knight 1998, Reiss 2000, Simpson 2000, Singleton et al. 2001). In the northwest Berkshires, there is a sharp ecotone between eastern deciduous and montane coniferous forests at approximately 760m. The red spruce-balsam fir forest dominates above this elevation. This study reached a maximum elevation of 740m and therefore focuses only on the eastern deciduous forest that lies below 760m (Braun 1950, Art and Dethier 1986, Barnes 1991). The landscape is a mosaic of forest stands of different ages and land-use histories. Stands that have developed in the past 50 years are predominantly a mix of grey birch (Betula populifolia), paper birch, quaking aspen (Populus tremuloides), and red maple. Red maple and red oak (Quercus rubra) tend to dominate stands 75 to 150 years old. Sugar maple and beech are the dominant species in primary woodlots and in secondary woodlots that are older than 150 years (Art and Dethier 1986). Disturbances are common in primary and maturing secondary woodlots, black birch and yellow birch (Betula lenta and B. alleghaniensis) are important colonizers of the light gaps that open between the canopy species (Braun 1950, Art and Dethier 1986). Certain primary woodlots are dense stands of eastern hemlock (Tsuga canadensis). With the exception of the occasional birch or red maple, hemlock stands are nearly monocultures (Smith 2005). In addition, the understorey of hemlock stands tends to be dark, acidic, low-nutrient places with a depauperate herbaceous layer (Barnes 1991). Land-use history The original and present day forest aspects bear little resemblance to one another. Three centuries of human occupancy have left their mark. Everywhere are seen the secondary communities which followed repeated cutting and burning of hardwood and mixed forest, and abandonment of farm and pasture land. Braun (1950, p.428) Humans are an important force in structuring the landscape. Between a third and a half of Earth's land area has been transformed for human use and this percent is growing (Vitousek et al. 1997). New England, opposite to the global trend, is in a period of reforestation after extensive agriculture and silviculture peaking in the mid 19~~ century (Singleton et al. 2001). Today New England's forests are a patchwork of old-woodlots and old-farmland. An important goal of ecology is to understand how human-uses alter the landscape and New England's mosaic of land-use histories provides an excellent opportunity to study these alterations in the long-term. Native Americans traveled through, but probably never permanently settled this part of the Berkshires. The European settlement of Williamstown started in the 1750s. Forest was cleared for agriculture at a rapid rate and by the 1830s, at least two-thirds of the land was clear of trees (Saterson 1977). In some places, land clearing reached the ridge crests of the Taconic Range and the Greylock Massif. Virtually all the forest left standing was selectively harvested for firewood and timber (Art and Dethier 1986). Thus, there are lands that have been continuously forested but no land has what Braun (1950) calls "virgin forest." The few sites in Williamstown that contain the oldest-growth trees are typically situated on steep and rocky slopes, consequently ill-suited for agriculture and difficult to log. In the 1830s, agriculture began to decline in the Berkshires. Much of the cultivated land became pastures for sheep and dairy farming. Eventually pastures also were abandoned and most of the once cleared land has since returned to forest (Art and Dethier 1986). Williamstown's land-use has turned the landscape into a patchwork of primary woodlots (logged but never cleared) and forests arising on abandoned agricultural lands (Saterson 1977). METHODS Study sites We sampled 77 plots: 54 primary deciduous, 17 secondary deciduous, and 6 primary hemlock. Through the course a decade, Henry Art and his thesis students established these study plots and surveyed the herbaceous and woody communities at each of these plots. They identified plants to species and determined tree basal area and herbaceous percent-cover. See Nash 1994, Grossmann 1996, Halbach 1997, Knight 1998, Reiss 2000, Simpson 2000, and McBride 2006 for exact vegetation sampling methods. The plots represent all the primary deciduous forest stands in Williamstown and a considerable number of the secondary deciduous and primary hemlock stands. They were selected somewhat systematically using the 1830s woodlot map of Williamstown, drawn at the height of deforestation (Figure 1,Mills 1830). For the purposes of this study, we separated forest stands into primary deciduous, primary hemlock, and secondary woodlots. Hemlock forests are have many unique characteristics and therefore deserve to be differentiated from primary deciduous. There are no secondary hemlock stands in Williamstown and consequently none are included in this study. p-y4y:... '. :.. . a. ..-.. $;i'.;.k;': '.' .".,i;;;,;i:g;:: .. ,-- h. .-- I?.: ! .AS. --3, 3':-:,.. , , -, >.. .. . ...-,_ 3: . :>I .. We used a Trimble TSCl GPS unit and Pathfinder Office 3.10 and ArcMap 9.1 GIs software to map our plots and measure elevation. We recorded aspect and slope using a Silva forestry compass and clinometer. The plots ranged in elevation from 170-740m, averaging about 375m. They covered all aspects and topographies, including ridges and ravines. The plots are circular with a radius of 16m and area of 804m2. Geographically, they are well distributed across Williamstown. There are plots on the Taconic, Greylock, and Green Mountain Ranges and all valleys between. One of our major goals was to represent the diversity of eastern deciduous-forest stands that Williamstown offers. Thus the results can have broader implications. Invertebrate sampling At each plot, we obtained 4 sampling points by randomly generating compass bearings and distances from center points. This sampling scheme alone would inevitably lead to clustering around the center point. To prevent this problem, we stratified the circular plot into 4 concentric rings of equal area, with a sampling point assigned to each ring. We used a 2Ox20cm quadrat frame and a garden trowel to take each sample. A corer would have been inappropriate in many cases because of the rochness of the soil. In each quadrat, we collected by hand the organic horizons (leaf litter, decomposition, humuslayers). Then using the trowel we cut out and removed 20x20x10cm of the A-horizon, the first mineral layer. To prevent invertebrate escapes, we accomplished these samplings with haste, scooping up organic layers and swiftly cutting out the mineral soil. Organic and mineral layers were stored in zip-closure bags at approximately 3°C and processed as soon as possible. In certain soils, the boundary between the organic horizons and the mineral horizons was ill-defined. We used our best judgment to estimate where the majority of the material changed from amorphous, decomposed leaf matter to mineral soil. We took samples during the day between 900 and 1500h and from mid June and early September. Our sampling schedule was governed by proximity of plots to trail heads, the weather, Vespula spp. attacks, and other field-work considerations. We spread the sampling of the different forest types throughout the summer to remove temporal biases. Invertebrate extraction We extracted invertebrates from organic and soil samples using an invertebrate extracting device modified from Wyman et al. (1999). Wyman's device is based on the same principles as the Tullgren-Berlese funnel, in which litter or soil samples are spread on a wire screen in a funnel. An incandescent light bulb suspended over the funnel drives invertebrates down through the litter, past the screen, and into a container of alcohol waiting below the funnel (Triplehorn and Johnson 2005). Wyman et al. (1 999) enclosed the system in foam-board insulation and installed an ice tray below the invertebrate collection point, creating a steep thermal gradient. A foam lid contains three 50 and three 75 watt incandescent bulbs and is sealed over the foam box containing samples to be extracted. Wyman also introduced a small fan, which further reduces moisture. The overall result is "a lighted area above the leaves that is hot and dry, and an area below the leaves that is dark, cool, and wet." (Wyman et al. 1999). By using a box instead of a funnel, Wyman increased the surface area through which invertebrates can exit the litter sample, making extraction more reliable. Wyman's extractors collect invertebrates in water, instead of alcohol. In these devices, alcohol vapors would discourage invertebrates from dropping out of the sample (Wyman et al. 1999). Samples in Tullgren-Berlese funnels are typically run for 3-4 days. Wyman et al.'s (1999) device reduces that time to 24h. These devices allow cheap, quick and relatively efficient extraction of invertebrates. Wyman reports extraction efficiencies of 50-loo%, differing taxon by taxon. Snails, for example, extract poorly. Instead of escaping downwards into the collection tray, snails retreat into their shells, where they desiccate. This is not a significant problem because snails are easily sorted from dried samples by hand. Desiccation resistant taxa also extract poorly. In certain cases, ants are able to withstand the heat and aridity that forces the more succulent taxa into the collection trays. See Wyman et al. (1999) for more information on these devices. We altered slightly Wyman et al. (1999)'s design to suit our purposes (Figure 2). Wyman et al. (1999) had used their device only for litter samples. Our project required extraction of invertebrates from both soil and litter samples. For ow soil devices, we decreased the number of holes in the support for the soil. This makes it slightly harder for invertebrates to escape, but we found that a high density of holes allows too much soil to fall into the invertebrate collection tray as it dries. In addition, we used aluminum trays as the invertebrate collection component, instead of following Wyman's design of calking a platform into the foam box. To facilitate removal of invertebrates from the aluminum trays, we installed a pouring funnel into the side of each collection tray. After extraction, invertebrates were simply poured out and into a polyethylene bottle. bvl spout Figure 2. Invertebrate extractors modified from Wyman et al. (1999), measuring approximately 120x30x15cm. The heat and light from the incandescent bulbs drives invertebrates down through the perforated foam layer into a water bath below. Drawn by Abby McBride. We hand searched dried soil and litter samples after they were run through the extractors. This improved the extraction efficiency of the larger individuals but was ineffective for smaller individuals, probably less than 5mrn in length. Invertebrate identification We sieved the invertebrate samples using a lmrn US standard 3" stainless-steel test sieve. Anything that passed through the lrnm sieve was discarded, greatly facilitating identification. For identification we spread invertebrate samples on a 9x9xlcm dish marked with a lcm grid, which helped ensure individuals were counted only once. Identifications were made using a dissecting microscope at 6x power unless an individual's characteristics were unclear, in which case we used as much as 50x magnification. We used Triplehorn and Johnson (2005) for most of the adult taxonomy and Chu and Cutkomp (1992) for the larval taxonomy. Kevan and Scudder (1989), Dinda1(1990), McGavin (2000), Gibb and Oseto (2006) also proved extremely useful. We identified all invertebrates to order. Ants were identified to family (Formicidae). Annelids were identified to potworms (Enchytraeidae) or earthworms (Lumbricidae). Gastropoda was identified to slugs or snails. We counted larvae separately, but did not count pupae because our extraction methods required active participation. To avoid counting decapitated individuals twice, we counted only heads. Soil chemistry and texture At each plot, we haphazardly took two lOcm deep, 6cm diameter soil cores from the upper A soil horizon. The two cores were placed in one zip-closure bag and mixed to obtain an average soil sample for each plot. These were stored at 3OC until processing. We measured pH following the dry-soil glass-electrode procedure of McLean (1982, 12- 2.6.5, p.208) and used pH 4.00 and 7.00 buffers for calibration. pH was read from soil- water slurries, and therefore the data are soil pH in water. For simplicity, these data will be referred to as soil pH. We extracted exchangeable cations from our soil samples using the Biichner funnel filtration, 1M ammonium acetate extraction procedure from Knudsen et al. (1982, 13- 3.3.1.3, p. 229). A flame atomic absorption spectrophotometer with direct aspiration measured the magnesium, calcium, potassium, and sodium cation concentrations of our soil extracts. We measured soil cation exchange capacity following Wilde et al. (1964). Cation exchange capacity measures the ability of soil to hold cations. Soil texture or particle-size analyses were carried out using standard hydrometer methods (Gee and Bauder 1986, 15-5.2). Data are expressed as percent soil content of sand, silt, and clay. Moisture was estimated using the Bormann Moisture Index, a 1-7 scale where 7 is the wettest. We measured soil organic matter content by loss-on-ignition (LOT). We weighed a dried soil sample and then ashed it in crucible in a 600°C muffle oven overnight. The sample was weighed again, allowing us to calculate %LOI. Statistical methods The data were loglo(x + 0.5) or square root transformed when they did not fit the normal distribution. These transformations are commonly used on hemiedaphic invertebrate data to improve homogeneity of variance (Williams and Gaston 1994, Liiri et al. 2002, Addison et al. 2003, Scheu et al. 2003, Hasegawa et al. in press). Analyses were then performed using JMP version 5.1 (2004). ANOVAs and Tukey's Honestly Significant Difference (HSD) post-hoc test were used at an a-level of 0.05 to compare means. Principal component analyses were also used to compare communities among forest types and with environmental variables. Linear regressions were used to build models to explain invertebrate density from a combination of variables, which were selected with a step-wise analysis. Only pertinent variables were included. mSULTS MD DISCUSSION Invertebrates A total of 9120 individual invertebrates were counted in 32 taxa, and an additional 3 larval groups. The mean site density of all invertebrates was 567.4 ind. m-2. Taxa abundances are summarized in Table 1. See Appendix for a table of invertebrates and their mean densities at each site. Invertebrate size (largest dimension) ranged from 2mm to approximately 100mm (earthworms). The community dominanceldiversity structure fits a log-normal model (R~adj.=0.98) (Figure 3). These data are underestimates of actual invertebrate densities. Invertebrate extraction with Wyman et al.'s (1999) devices is more efficient than Tullgren-Berlese funnels, but still has significant margin of error. Because the majority of soil and litter invertebrate studies use Tullgren-Berlese funnels, it is difficult to compare these data to the literature until more efficient extraction methods become prevalent. In addition, it is possible that more mobile invertebrates avoided capture during litter and soil collection, thus under- representing mobile groups. The methods were consistent between the 77 sites and therefore comparisons between the sites can be made with confidence. Table 1. Invertebrates (22mm) densities among all 77 sites. Taxon sum of means mean of means Collembola Julida Coleopteva (larvae) Diptera (larvae) Araneida Formicidae Coleoptera (adults) Pseudoscorpionida Geophilomorpha Chovdeumida Acavi Lumbricidae Enchytvae idae Lepldoptera (larvae) Lithobiomorpha Hemiptera Gastropods (snails) Polydesmida Hymenoptera (adults) Psocoptera Symphyla Diptera (adults) Isopoda Thysanoptera Nematoda Gastropods (slugs) Opiliones Orthopteva Hymenoptera (larvae) Polyzoniida Dermaptera Polyxenida Neuroptera (larvae) Mecopteva (larvae) Diplura Ind. m2 567.4 16.2 84.77 75.81 66.77 5 1.68 47.40 42.32 33.60 29.30 25.08 16.64 16.56 13.39 10.55 10.39 9.58 6.25 5.03 4.95 3.73 3.00 2.35 2.11 1.38 1.38 1.06 0.73 0.41 0.32 0.24 0.16 0.16 0.08 0.08 0.08 0.08 Notes density of all invertebrates 22mm density of the mean invertebrate taxon Hexapoda <5mm cylinder millipedes (Diplopoda) beetle larvae fly larvae spiders ants (family) beetles arachnid <5mm centipede (Chilopoda) millipede mites (Arachnida) earthworms (Oligochaeta) potworms (Oligochaeta) caterpillars, no adults encountered centipede true bugs (Hornoptera included) gastropods with developed shells flat-backed millipedes adults excluding Formicidae barklice and booklice Myriapodai5mm adult flies pill bugs (Crustacea) thrips <4mm roundworms <4mm gastropods without shells harvestmen (Avachnida) crickets and grasshoppers larvae excluding Fovmicidae millipedes earwigs bristly millipedes <4mm no adults encountered no adults encountered Hexapoda -s I I I I I I I 0 5 I0 15 20 25 30 35 Rank he rank-abundance curve (blue line) is a log transformed regression (R~ adj.=0.98, P 5% of the variation are similarly non-significant (not shown). Indicator taxa PCA A third way to quantify a community is to pick ecologically important taxa. This approach assumes that other taxa are not significant in community composition and structure and moreover that these taxa are obscuring potential community differences in the important taxa. In these communities the indicator taxa were springtails, beetle larvae,Julida millipedes, spiders, fly larvae, and Oligochaetes (potworms and earthworms). These taxa play important community roles in their ecological fbnctions and large abundances. The indicator taxa PCA required 3 PCs to account for greater than 70% of the indicator taxa variation, showing that there was less variation among indicator taxa than the whole community (which required 14 PCs). Nevertheless, none of the indicator taxa PCs significantly differed among forest types (Figures 7). There are three main possibilities. First, there are no community differences among the forest types. Second, the chosen indicator taxa may be poor community indicators. Third, the taxonomic level may be too high to detect community differences (see Community differences). 3 2- T a 1- 0- C c 0 .-& 53 .- -1--2-: --;--4 -0 c- -3- -4 -5 I I Hemlock Primary Secondary Forest type of the total indicator taxa variation and does not differ significantly among forest types. The rest of the indicator taxa principal components are similarly non-significant. Functional groups Functional groups are a fourth way to categorize a community, however at the order taxonomic level it is difficult to group taxa into functional groups. For example, the order Coleoptera includes herbivores, detritivores, fungivores, and carnivores (Dindal 1990). At order level, it is possible to distinguish only two functional groups: detritivores (in a broad sense) and predators. Diverse orders like Coleoptera must be left out of this analysis. Detritivores include springtails, millipedes, isopods, Oligochaetes, and Gastropods (slugs and snails). Predators include spiders, harvestmen (Opiliones), pseudoscorpions, and centipedes. The rest of the taxa were excluded. Neither predator abundance nor detritivore abundance differed significantly among forest types (Figure 8). 300-250 'u!100 a P 0 & b- 50- 0 Hemlock I Pr~mary I Secondary I Hemlock Primary Secondary I a> Forest type Forest type Figure 8. a) Predator density in the three forest types. The means are not significantly different. b) Detritivore density in the three forest types. The means are not significantly different with or without the outliers present. Community dzflerences The four community measures (a-diversity, community PCA, indicator PCA, and hnctional groups) failed to detect differences among forest types. At the order taxonomic level these communities are not significantly different, but orders include diverse species and genera with many different of ecological roles and physiological limitations. Orders and families like Diptera, Coleoptera, Collembola, and Formicidae are abundant across the world from high latitudes to the equator. It is not surprising that the orderlfamily resolution does not reveal community differences across a spatial scale of only tens of lulometers. The top predator The top of the trophic web is an important group to examine in any community. The size of the upper trophic level is often indicative of community assembly that is more difficult to discern at lower trophic levels. Spiders are important in all terrestrial ecosystems and are the top predators in the hemiedaphic invertebrate system (Isaia et al. 2006). Despite the lack of overall community differences, spider densities differed significantly among forest types (P =0.025). Mean spider density was 54.7k9.7 ind. m-2 in primary woodlots, 31.6317.3 ind. m-2in secondary woodlots, and 26.0t29.2 ind. m-2 in hemlock forests (Figure 9). 200 150-i 03 g 100-50--------. :s o*+ 0- I 1 Hemlock Primary Secondary Forest type Figure 9. Spider (Araneida) density by forest type. Mean diamonds represent group means and 95% confidence interval. These data are snrnewh~~t cke~wellin the llnnpr rslnop --------. -----------------'---A-..A*-.. UILV 1. "U AIA CAI" -rrwA 6' but the means are significantly different (P =0.025, ANOVA). Variation in the hemlock spider density was large and consequently non-significant. Hemlock stands create strilangly different habitat than do deciduous forests and a sample size of 6 hemlock stands is not adequate to understand the complexities of these environments. Hemlock stands are more ecologically similar to Douglas-fir forests than the deciduous forests in this study. Trofymow et al. (2003) find that spider communities decrease in abundance and complexity through succession in a Canadian Douglas-fir forest, potentially due to the homogeneity of old coniferous stands. This contrasts with the findings of studies in other forest types (Willet 2001 and Wheater et al. 2000), but explains why in the present study (though non-significant) the hemlock spider density was lower than the deciduous primary and secondary spider density. This study had a low sample size of primary hemlock stands and moreover there were no available secondary-hemlock stands for comparison. Hemlock plot data are absent from the rest of these analyses. Spider density did not adequately fit the normal curve, an ANOVA assumption. As expected with ecological systems, the data were skewed to the high densities. A log transformation did not properly normalize the data, but a square root transformation met the assumptions (Figure 10a). Square root adjusted spider density confirms that primary plots have a significantly higher spider density than secondary plots (Figure 1 Ob). This is significant according to Student's t at P <0.05. 14 12-i 10--a .- E a-6- CI 0 2 4- 0- 2-0--2 I Primary Secondary b) Forest type Figure 10. a) Histogram of square root transformed spider density meeting the normal assumptions. X-axis is sq root of spider density and y-axis is counts at those densities. b) Square root transformed spider density by forest type. These data fit the normal curve and the means are still significantly different (R~adj.=0.042, P =0.048). Influences on spider density Depth of the leaf litter layer is significantly, positively correlated with spider density (R~ adj.=O. 12, P <0.005). When both forest type and litter depth are used in a linear regression model to explain spider density, the model is significant (Table 6 and Figure 11). Leaf litter is vital in these communities because it is both the habitat space and food resource. In this model forest type plays an influential but non-significant role, suggesting that the model does not fully reflect the complexity of the system. The model in Table 6 and Figure 1 1 has leaf litter influence spider density equivalently in primary and secondary woodlots. The difference between primary and secondary woodlots is reflected in the y-intercept of each regression (Figure 11). Table 6. Summary of spider density linear regression model shown in Figure 11. Estimates indicate the magnitude and direction of correlation. P-value Variable Estimate Std error R~adj. (ANOVA) --0.147 whole model 0.0017 - forest type (1") 0.657 0.357 0.0699 litter 2.201 0.714 -0.0030 Pigure I I. apiaer uensiiy rlrlt;il;r l~gl~331ull N ~ I~l++++u+r ylv Lo llu II1lbIIVI U L. ~ mean than secondary plots regardless of leaf litter depth. The slopes of the primary and secondary regressions are equal. The slope is determined by a regression of litter depth by spider density. The difference between the primary and secondary regression is their y-intercepts. These are determined by the mean spider densities in primary and secondary plots. A step-wise linear-regression model including all pertinent environmental and floristic variables produced the model summarized in Table 7. These variables have potential influences on spider density, but this model is not entirely satisfactory. Many of the variables do not fit well in the model (P>0.08). Forest type again is the second most influential variable but is non-significant. Moisture is nearly as influential as forest type but is significant. Other potentially meaningful variables include soil organic content, K+ concentrations, and Acer saccharurn abundance. Table 7. Summary of spider density linear regression model. This model included all pertinent environmental and floristic variables and was completed step-wise. P-value Variable Estimate Std error R~adj. (ANOVA) whole model 0.3 17 0.0009 forest type (1") 0.749 0.462 -0.1115 litter 2.460 0.81 1 -0.0039 moisture 0.723 0.289 0.0158 oranic content 0.040 0.025 0.1 173 ca2' 0.000 0.000 0.1997 K' -0.020 0.012 0.0843 tree a-diversity 0.216 0.164 0.1934 - Acer rubrzrrn -0.022 0.017 0.2199 Acer saccharurn -0.028 0.012 -0.0268 Tszlga and Pinus -0.047 0.026 -0.0847 The previous regression analyses assume that variables affect spider density independently of each other. In the case of most variables, this is true. For example, spider density has a weak, negative correlation with K' concentrations in both primary and secondary woodlots. The relationship is the same no matter the land-use history. Leaf litter, on the other hand, does not meet this assumption. Spider density has a different relationship with litter in primary and secondary plots. In primary plots, spider density increases with litter depth (R~adj.=O. 18, P <0.001). Spider density has no significant relationship with the depth of secondary woodlot leaf litter (R~adj.=0.0). This indicates that either primary leaf litter is different than secondary leaf litter or conditions in primary woodlots are such that the leaf litter resource is more accessible to spiders and spider prey. The difference could be increased habitat space, increased food resource, or both. Because leaf litter serves both as habitat space and food resource in these systems, it is difficult to discern which is more important (Scheu et al. 2003). The differing effects of primary and secondary leaf litter require a model that allows leaf litter to function dependent on forest type. The regression model summarized in Table 8 more accurately represents the relationship between litter, forest type, and spider abundance. This model allows litter to act differently in primary and secondary woodlots. This dependent measure is termed "forest type (lo)x litter." In this model, forest type regains its statistical significance. Litter is not significant in the whole model because it is a function of forest type. The action of litter dependent of forest type, however, is significant and furthermore is the most influential variable in the model. These trends are represented graphically in Figure 12 using the model summarized in Table 8. Table 8. Summary of spider density linear regression model. This model assumes that litter affects spider density differently in primary and secondary plots. "Forest type (1O) x litter" is the action of litter dependent of forest type. P-value Variable Estimate Std error R~adj. (ANOVA) whole model - - 0.187 0.0007 forest type (1") litter 0.751 1.086 0.351 0.878 -- 0.0361 0.2206 forest type (1") x litter 1.832 0.878 0.0407 Figure 12. Spider density linear regression model plot from model sumrnarized in Table 8. Spider density increases with leaf litter in primary woodlots but not in secondary woodlots. a) The correlation in primary woodlots is significant (R2adj. =O. 1 8, P <0.00 I), but the relationship in secondary woodlots is non-significant (R2adj. =0.0). b) The model remains significant with the two lower outliers removed (R~adj. =O. 153 P =O. 003). The model in Table 8 and Figure 12 is an extreme simplification. Spider abundance is governed by more than just forest type and litter amount. Of the environmental variables measured in this study, elevation and moisture play the most important roles in determining spider density. Coniferous trees (Tsuga canadensis and Pinus spp.) and Acer saccharurn also seem to influence spider communities, possibly through leaf or needle litter inputs. A linear regression model of spider density which used a step-wise analysis to select from pertinent environmental and arboreal variables is summarized in Table 9. Table 9. Summary of spider density linear regression model. This model assumes that leaf litter affects spider density differently in primary and secondary woodlots. P-value Variable Estimate Std error R~adj. (ANOVA) whole model --0.287 0.0003 forest type (1 ") 1.019 0.393 -0.0121 litter 1.856 0.891 0.0417 forest type (I ") x litter 2.039 0.851 0.0198 elevation 0.004 0.002 0.0692 moisture 0.606 0.278 0.0335 Acer saccharzlm -0.033 0.01 1 -0.0044 Tsugaand Pinus -0.032 0.025 -0.2178 Unfortunately, leaf litter depth was the only leaf litter parameter that this study measured. Litter depth is important because it represents habitat space but it does not account for resource quality. The other environmental variables measurements used soil samples, not the litter layer. Soil may be more representative of bedrock than leaf litter conditions. There was no clear link among the soil environmental variables and the difference in primary and secondary leaf litter. Tree species are obviously linked to litter layer composition, but again there was no clear difference in tree species. Acer saccharum was more abundant in primary woodlots, but the model in Table 9 shows that spiders have a slight negative correlation with that species. In a single linear regression, there is no correlation between Acer saccharum and spider density (R~adj.=O.O). In fact, spider density does not significantly correlate with any of the important canopy species. Yet, the fit that trees have in the linear regression models suggests that a connection between spiders, litter, and canopy composition may exist. This study did not measure the correct leaf litter parameters to draw any such conclusions. To better understand the connections between canopy composition and the hemiedaphic invertebrate community, future studies should quantify the litter inputs from canopy trees and measure the chemical properties of the litter layer. Community implications Willet (2001) and Wheater et al. (2000) used spiders as indicators of logging intensity, forest succession, and restoration. Although Willet (2001) was looking at redwood forests and Wheater et al. (2000) were studying reclaimed limestone quarries, their findings can be applied to land-use history in eastern deciduous forests. Willet (2001) explains that litter spiders are particularly good community indicators because they are "directly affected by habitat structure and prey abundance." Spiders are sensitive also to "a wide range of environmental factors" and "as top predators in invertebrate food chains, some spiders may be sensitive to the structure and development of prey communities." (Wheater et al. 2000). Both Willet (2001) and Wheater et al. (2000) find that spider community abundance and complexity increases with succession and restoration. In particular, Willet (2001) explains that "spiders reacted favorably to widely spaced overstory trees," typical of older forests. The present study shows that spiders also serve as indicators of land-use history in eastern deciduous forests, even after as much as 150 years of natural succession. The lower abundance of spiders in secondary woodlots suggests that agriculture homogenizes the hemiedaphic system and reduces availability of prey species large enough to support the top invertebrate predator, spiders. Changes in the spider community indicate changes in prey groups. The survey of the rest of the community was at too high a taxonomic level to resolve these differences. Prey abundance could change without noticeably changing order level abundances if prey species were replaced by non-palatable species of the same order. Leaf litter is important as habitat space for spiders and prey and is the food resource for the entire hemiedaphic food web. Primary woodlots appear to have higher quality leaf litter than do secondary woodlots. Why and how primary woodlot litter is of higher quality is unclear because this study did not directly measure litter properties. Primary woodlot litter may influence spider density as a food resource in a bottom-up interaction through the food web. Additionally, litter directly affects spider and prey density by providing habitat space. A detailed study of litter composition is necessary to fully understand these processes. Regression models show that moisture is the next most important factor in determining spider density after litter depth and forest type. Primary woodlots are nearly significantly more moist than secondary woodlots (P=0.0695) and spider abundance increases both with moisture and in primary woodlots. Spiders are weakly sclerotized and are susceptible to desiccation (Jabin et al. 2004). In addition, species adapted to drier soil conditions have more sclerotized exoskeletons and may be less palatable to spider predators (Convey 2003). The regression model shows that canopy composition plays an important role in community assembly, but it is unclear how this functions. Most likely canopy composition is related to litter composition. Environmental influences Spiders as community indicators are influenced by both land-use history and environmental variables. The overall community measures (a-diversity, community PCA, etc.) did not differ by land-use. Instead, environmental and arboreal variables explain a significant portion of the variation in these community measures. Many of these variables are highly interrelated. For example, nutrients, pH, and moisture are well correlated with elevation (see elevational gradients section). As a result, it is easy to notice that a group of environmental factors are influencing a community, but without detailed study it is impossible to tell which forces are directly affecting the distribution and abundance of invertebrates. a-diversity The number of taxa in a community is well influenced by the elevation-pH-nutrient complex. These variables influence the hemiedaphic environment in concert. Three plausible linear regression models of a-diversity are shown in Tables 10-12 and explain approximately 20% of the variation each. These variables must be included in models separately because they are so highly correlated. Therefore three models accounting for 20% do not add up to 60% of the total variation. They merely demonstrate that there are a number of possibilities which require further study to understand. The negative correlation between beech abundance and a-diversity (Table 10 and 11) can be explained a few ways. Beech abundance increases with elevation (R~ adj.=0.29, P <0.001) and acidity (R~adj.=0.045, P <0.05). Beech may grow preferentially on acidic soils or beech may actively acidify the soils. Art and Dethier (1986) found that mature beech stands acidify precipitation more than mature sugar maple stands. The acidification may be due to canopy or litter processes. Availability of ca2+ and M~~+ decreases rapidly at low pH, even over small increments (Berg and Hemerik 2004). In addition, Berg and Hemerik (2004) explain that "most macroinvertebrate species are known to be acid sensitive." Beech litter may be of poor quality for the decomposer community. Fagaceous litter has nearly half the ash content of non-fagaceous angiosperms. Low ash content is characteristic of forest sites of low fertility, which can be due to acidity or simply low nutrient concentrations (Jensen 1974). McBride (2006) explains that forest herb cover is reduced when beech is abundant because its litter acts as a physical or chemical barrier to growth. The beech litter properties affecting the herbaceous community may be influencing the invertebrates also. Additionally, the herb community may be a responding to the reduction in the invertebrate decomposer community because of reduced nutrient cycling. Table 10. Summary of invertebrate a-diversity linear regression model. P-value Variable Estimate Std error R2 adi. (ANOVA) whole model --0.200 0.0002 soil pH 0.863 0.351 -0.0166 beech abundance -0.031 0.010 -0 0042 Table 11. Summary of invertebrate a-diversity linear regression model. P-value Variable Estimate Std error R2adj. (ANOVA) whole model --0.172 0.0007 Mg2+ 0.002 0.001 -0.0704 beech abundance -0.031 0.010 -0.0006 Table 12. Summary of invertebrate a-diversity linear regression model. P-value Variable Estimate Std error R2adj. (ANOVA) whole model --0.178 0.0005 elevation -0.005 0.002 -0.0113 soil pH 0.799 0.366 -0.0327 Community PCA The community principal component 1 (PCI, Figure 6a) correlates with several important environmental variables. This demonstrates that the environmental variables measured in this study are significant in influencing the invertebrate community as a whole. Community PC1 is positively correlated with millipedes, mites, isopods, et al., but negatively correlated with spiders, springtails, beetle larvae, fly larvae, et al. The model summarized in Table 13 shows that soil moisture, ~g~', and litter depth influence community PC1. This model supports the importance of the leaf litter resource for these invertebrate communities, but certain taxa are more abundant at sites with smaller litter depths. According to the model in Table 13, these taxa are relatively desiccation- resistant and may thrive in the drier environments of smaller litter layers. This model illustrates the negative relationship between litter thickness and moisture. Thick litter layers trap moisture and the model shows that hydrophilic species like fly larvae prefer thick litter layers. Table 13. Sunmary of community PC1 linear regression model. P-value Variable Estimate Std error R~adj. (ANOVA) whole model --0.276 <0.0001 moisture 0.432 0.164 -0.0104 Mg2t 0.003 0.001 -0.0021 litter -1.079 0.483 -0.0290 Moisture is an important physiological limitation for hemiedaphic invertebrates. Species can invest in hard outer parts to preserve water but this comes at a cost. For example, millipedes, certain mites, and isopods are highly dependent on ca2' concentrations to build the sclerotized exoskeletons that protect them from desiccation (Convey et al. 2003, Berg and Hemerik 2004). In this study, Julida millipede densities increase with ca2+ (R~ adj.=0.24,P <0.0001) and pH (R~ adj.=0.20,P <0.0001). Community and environmental PCAs PCA of environmental variables provides a further demonstration of the relationships between hemiedaphic invertebrate communities and the environmental variables measured in this study. PCA analysis combined variation across the many abiotic variables into 14 axes. The best 5 of these are required to account for >70% of the total variation in abiotic variables. A step-wise linear regression model of community PC1 using the 5 abiotic PCs produced the model summarized in Table 14. Abiotic PC1 and PC3 account for 32.9% of the variation in community PC1. Abiotic PC1 represents a decrease in elevation and an increase in pH, moisture, and nutrient concentrations. Abiotic PC2 follows an increase in elevation and a decrease in nutrient concentrations. Tablc 14. Sulmna~y of'co~mnunity PC1 linear regression modcl. P-value Variable Estimate Std error R~adj. (ANOVA) whole model --0.329 <0.0001 Abiotic PC 1 0.496 0.104 -<0.0001 Abiotic PC3 -0.406 0.146 -0.0073 Overall, the community measures are influenced by a similar suite of environmental variables as spider density. This agreement supports the notion of spiders as community indicators. The elevation, pH, ca2+, ~g~', moisture, and beech abundance complex of environmental variation is extremely important in structuring the hemiedaphic invertebrate communities of eastern deciduous-forest floors. Litter quantity and quality also are vital for these communities but are a function of canopy composition. These parameters correlate with elevation when canopy species like beech have elevational preferences. Variance The spider density, a-diversity, and PCA linear regression models account for at most a third of the total variation. This indicates that while the trends are significant, they do not represent the whole picture. The unrepresented variation can come from several sources. First, there must be a significant amount of random stochasticity in the system. Random environmental events influence the distribution and abundance of all organisms, including those in the hemiedaphic system. These variations cannot be explained by land-use of the environmental variables measured in this study. Second, invertebrate distribution and abundance changes temporally. At the three repeatedly sampled plots, densities of important taxa were found to vary significantly through the course of the summer (Figures 13-15). These changes can explain the variation not accounted for by the variables sampled in this study. Third, the sampling methods in this study have significant margins of error. Wyman et al. (1999) estimate the success of the invertebrate extractors as upwards of 50%. If the error rate varied from sample to sample, it would account for the large overall variation. Figure 13. Temporal variation in springtail density (ind. m-2) at three sites sampled successively through the summer. Blue is Pine Cobble Lower, a primary stand. Pink is Phelps Knoll, a secondary stand. Yellow is Torrey Woods, a hemlock stand. Figure 14. Temporal variation in Julida millipede density. Colors represent the same sites same as in the previous figure. Figure 15. Temporal variation in spider density. Colors represent the same sites same as in the previous figure. These forces of variation do not detract meaning from the results of the present study. The 30% of community variation accounted for by the models in this study is a significant portion. Moreover, this 30% may represent the important differences between communities. The unaccounted for 70% may be relatively unimportant in the overall structure and function of these communities. Spiders for example are a numerically only 8% of the mean invertebrate density, but are far more important members of the hemiedaphic food web than this number represents. CONCLUSIONS 1. Spider densities indicate that post-agricultural eastern deciduous-forests support different hemiedaphic-invertebrate communities than do primary eastern deciduous-forests. Leaf litter quality in primary woodlots promotes a forest-floor habitat more suitable for spiders and their prey than litter in secondary woodlots. This is significant because spiders are the top invertebrate-predators in the decomposer food-web and may indirectly encourage rates of decomposition through trophic interactions. 2. A suite of interacting environmental variables influence the distribution and abundance of hemiedaphic invertebrates. Topography and elevation drives this complex, which includes pH, Ca2+, ~g~', moisture, and beech abundance K', (litter quality). 3. If spiders are indicative of community complexity, then it increases with moisture and litter quality. Community PCA suggests that certain invertebrate orders have special physiological requirements. Millipedes (Diplopoda) require high nutrient concentrations, to build their sclerotized exoskeletons. 4. Unless spiders are used as indicators, the order taxonomic level is not fine enough to resolve overall community differences among land-use histories. Most orders, especially the abundant ones (e.g., Coleopteva, Dipteva, Collembola) contain an array of species and genera with diverse physiological requirements. Furthermore, functionally distinct taxa within the same order can replace one another without changing the overall community at the order taxonomic resolution. 5. To better understand community differences between primary and secondary woodlots, future studies should pick ecologically important orders (e.g., Coleoptera)and identify to a lower taxonomic level, ideally genus or species. Although family may suffice. Aeknowliedgements I extend my gratitude to Hank Art. Working with him is a pleasure that will be missed. I thank Abby McBride and Jim Marlowe for their help in the field and lab. Thank you Manuel Morales for commenting on my draft. Bernhard Klingenberg, Manuel Morales, and David Smith generously provided many hours of statistics lessons. I am thankful for Jay Racela's constant help in the envi lab. ID critters day, conceived of and organized by Abby, allowed the completion of this project. Thank you Avery R. Briggs, Ellen Crocker, Tyler Corson-fikert, Clara Hard, Karl Naden, Carolyn Reuman, James fitterpusch, and Lisetta Shah. Larry Mattison, Emille Ouellete, and George Walther helped me make the invertebrate extractors and various other contraptions into works of art. Rick Wyman, inventor of the extractors, graciously let me into his lab to see them in action. This project was funded in part by gummy worms from my parents. References Addison, J.A., Trofymow, J.A., & Marshall, V.G. (2003) Abundance, species diversity, and community structure of Collembola in successional coastal temperate forests on Vancouver Island, Canada. Applied Soil Ecology, 24,233-246. Adl, S.M. (2003) The Ecology of Soil Decomposition CAB1 Publishing, Cambridge. Anderson, J.M. (1975) Succession, Diversity and Trophic Relationships of Some Soil Animals in Decomposing Leaf Litter. The Journal of Animal Ecology, 44,475-495. Andre, H.M., Ducarme, X., & Lebrum, P. (2002) Soil biodiversity: myth, reality or conning? Oikos, 96, 3-24. Art, H.W. & Dethier, D.P. (1986). Influence of Vegetative Succession on Soil Chemistry of the Berkshires, Rep. No. 153. University of Massachusetts Water Resources Research Center, Amherst. Bardgett, R.D. (2005) The Biology of Soil: A Community and Ecosystem Approach Oxford University Press, Oxford. Barnes, B.V. (1991). Deciduous Forests of North America. In Temperate Deciduous Forests (eds E. Rohrig & B. Ulrich), Vol. 7. Elsevier, New York. Barros, E., Pashanasi, B., Constantino, R., & Lavelle, P. (2002) Effects of land-use system on the soil macrofauna in western Brazilian Amazonia. Biology and Fertility of Soils, 35, 338-347. Berg, M.P. & Hemerik, L. (2004) Secondary succession of terrestrial isopod, centipede, and millipede communities in grasslands under restoration. Biology and Fertility of Soils, 40, 163-170. Bird, S.B., Coulson, R.N., & Fisher, R.F. (2004) Changes in soil and litter arthropod abundance following tree harvesting and site preparation in a loblolly pine (Pinus taeda L.) plantation. Forest Ecology and Management, 202, 195-208. Blair, J.M., Parmelee, R.W., & Wyman, R.L. (1994) A comparison of the forest floor invertebrate communities of four forest types in the northeastern U.S. Pedobiologia, 38, 146-160. Bloemers, G.F., Hodda, M., Lambshead, P.J.D., Lawton, J.H., & Wanless, F.R. (1 997) The effects of forest disturbance on diversity of tropical soil nematodes. Oecologia, 11 1, 575-582. Boonstra, F. (2004) Assessing sugar maple (Acer saccharum) decline in the Hophns Memorial Forest, Williams College, Williamstown. Braun, E.L. (1 950) Deciduous Forests of Eastern North America Hafner Publishing Company, New York. Burghouts, T., Ernsting, G., Korthals, G., & Vries, T.D. (1992) Litterfall, Leaf Litter Decomposition and Litter Invertebrates in Primary and Selectively Logged Dipterocarp Forest in Sabah, Malaysia. Philosophical Transactions: Biological Sciences, 335,407- 416. Chu, H.F. & Cutkomp, L.K. (1992) How to Know the Immature Insects, 2 edn. WCB McGraw-Hill, Boston. Convey, P., Block, W., & Peat, H.J. (2003) Soil arthropods as indicators of water stress in Antarctic terrestrial habitats? Global Change Biology, 9, 17 18- 1730. Deleporte, S. & Tillier, P. (1999) Long-tern effects of mineral amendments on soil fauna and humus in an acid beech forest floor. Forest Ecology and Management, 11 8,245-252. Dindal, D.L., ed. (1990) Soil Biology Guide. John Wiley & Sons, New York. Edwards, C.A. & Bohlen, P.J. (1996) Biology and Ecology of Earthworms, 3 edn. Chapman and Hall, London. Eggleton, P., Bignell, D.E., Hauser, S., Dibog, L., Norgrove, L., & Madong, B. (2002) Termite diversity across an anthropogenic disturbance gradient in the humid forest zone of West Africa. Agriculture, Ecosystems and Environment, 90, 189-202. Gee, G.W. & Bauder, J.W. (1986). Particle-size Analysis. In Methods of Soil Analysis (part 1) (ed A. Klute), Vol. 9. American Society of Agronomy and Soil Science Society of America, Madison. Gibb, T.J. & Oseto, C.Y. (2006) Arthropod Collection and Identification: Field and Laboratory Techniques Elsevier Academic Press, Boston. Gonzales, G. & Seastedt, T.R. (2001) Soil fauna and plant litter decomposition in tropical and subalpine forests. Ecology, 82, 955-964. Greenberg, C.H. & McGrane, A. (1996) A compasison of relative abundance and biomass of ground-dwelling arthropods under different forest management practices. Forest Ecology and Management, 89,3 1-41. Grossmann, E.B. (1996) Environmental mechanisms through which past land-use influences spring ephemeral wildflower distribution patterns in the Hopkins Memorial Forest, Williams College, Williamstown. Hairston, N.G., Smith, F.E., & Slobodkin, L.B. (1960) Community structure, population control, and competition. The American Naturalist, 94, 42 1-425. Halbach, S.M. (1997) The distribution of herbaceous species in primary and secondary woodlots in Williamstown, Williams College, Williamstown. Hasegawa, M., Fukuyama, K., Makino, S.i., Okochi, I., Goto, H., Mizoguchi, T., Sakata, T., & Tanaka, H. (in press, 2006) Collembolan community dynamics during deciduous forests regeneration in Japan. Pedobiologia, (in press). Hibbs, D.E. (1983) Forty Years of Forest Succession in Central New England. Ecology, 64, 1394-1401. Isaia, M., Bona, F., & Badino, G. (2006) Influence of Landscape Diversity and Agricultural Practices on Spider Assemblage in Italian Vineyards of Langa Astigiana (Northwest Italy). Environmental Entomology, 35,297-307. Jabin, M., Mohr, D., Kappes, H., & Topp, W. (2004) Influence of deadwood on density of soil macro-arthropods in a managed oak-beech forest. Forest Ecology and Management, 194,61-69. Jensen, V. (1974). Decomposition of Angiosperm Tree Leaf Litter. In Biology of Plant Litter Decomposition (eds C.H. Dickinson & G.J.F. Pugh), Vol. 1. Academic Press, New York. Kevan, D .K.M. & Scudder, G.C.E. (1989) Myriapods Biological Survey of Canada, Ottawa. Knight, J. (1998) Dispersal patterns of wildflowers from primary to secondary woodlots, Williams College, Williamstown. Knudsen, D., Peterson, G.A., & Pratt, P.F. (1982). Lithium, Sodium, and Potassium. In Methods of Soil Analysis (part 2) (ed A.L. Page), Vol. 9. American Society of Agronomy and Soil Science Society of America, Madison. Kounda-Gki, C., Vaqulik, A., Ponge, J.F., & Sarthou, C. (2004) Soil arthropods in a developmental succession on the Nouragues inselberg (French Guiana). Biology and Fertility of Soils, 40, 1 19-127. Liiri, M., Setala, H., Haimi, J., Pennanen, T., & Fritze, H. (2002) Relationship between soil microarthropod species diversity and plant growth does not change when the system is disturbed. Oikos, 96, 137-149. Lutz, H.J. & Chandler, R.F. (1966) Forest Soils John Wiley & Sons, New York. McBrayer, J.F., Ferris, J.M., Ferris, L.J., Metz, L.J., Gist, C.S., Cornaby, B.W., IOtazawa, Y., atazawa, T., Wernz, J.G., Krantz, G.W., & Jensen, H. (1977) Decomposer invertebrate populations in U.S. forest biomes. Pedobiologia, 17, 89-96. McBride, A. (2006) A biogeographical study of herbaceous plants in the forests of Williamstown, MA, Williams College, Williamstown. McGavin, G.C. (2000) Insects, spiders, and other terrestrial arthropods Dorling Kindersley, New York. McLean, E.O. (1982). Soil pH and Lime Requirement. In Methods of Soil Analysis (part 2) (ed A.L. Page), Vol. 9. American Society of Agronomy and Soil Science Society of America, Madison. Mikola, J. & Setala, H. (1998) No evidence of trophic cascades in experimental microbial-based soil food web. Ecology, 79, 153-164. Milcu, A., Partsch, S., Langel, R., & Scheu, S. (2006) The response of decomposers (earthworms, springtails and microorganisms) to variations in species and functional group diversity of plants. Oikos, 1 12,5 13-524. Mills, J. (1830) Williamstown Woodlot Map. Provided by Henry Art. Nash, K.M. (1994) The role of dispersal mechanisms on the colonization of spring herbaceous flora in secondary woodlands, Williams College, Williamstown. Neave, P. & Fox, C.A. (1998) Response of soil invertebrates to reduced tillage systems established on a clay loam soil. Applied Soil Ecology, 9,423-428. Niemela, J. (1997) Invertebrates and Boreal Forest Management. Conservation Biology, 11,601-610. Paquin, P. & Coderre, D. (1997) Changes in soil macroarthropod communities in relation to forest maturation through three successional stages in the Canadian boreal forest. Oecologia, 112, 104- 111. Ponsard, S., Arditi, R., & Jost, C. (2000) Assessing top-down and bottom-up control in a litter-based soil macroinvertebrate food chain. Oikos, 89, 524-540. Raty, M. & Huhta, V. (2004) Communities of Enchytraeidae (Oligochaeta) in planted birch stands as compared with natural forests in central Finland. Biology and Fertility of Soils, 40,413-420. Reiss, A.E. (2000) Influences on Distribution of Spring Ephemerals on the Mount Greylock Reservation, Williamstown, MA, Williams College, Williamstown. Santos, P.F., Phillips, J., & Whitford, W.G. (1981) The role of mites and nematodes in early stages of buried litter decomposition in a desert. Ecology, 62, 664-669. Saterson, K.A. (1977) A Vegetation History of Williamstown, Williams College, Williamstown. Schaefer, M. & Schauermann, J. (1990) The soil fauna of beech forests: comparison between a mull and a moder soil. Pedobiologia, 34,299-3 14. Scheu, S., Albers, D., Alphei, J., Buryn, R., Klages, U., Migge, S., Platner, C., & Salamon, J.-A. (2003) The soil fauna community in pure and mixed stands of beech and spruce of different age: trophic structure and structuring forces. Oikos, 10 1,225-23 8. Seastedt, T.R. & Crossley, D.A. (1981) Microarthropod Response Following Cable Logging and Clear-Cutting in the Southern Appalachians. Ecology, 62, 126- 13 5. Simpson, E.L.H. (2000) Factors Affecting the Distribution of Spring Ephemerals on the Taconic Range, Williams College, Williamstown. Singleton, R., Gardescu, S., Marks, P.L., & Geber, M.A. (2001) Forest Herb Colonization of Postagricultural Forests in Central New York State, USA. The Journal of Ecology, 89, 325-338. Smith, J. (2005) The Current Ecology of Eastern Hemlock (Tsuga canadensis) in Williamstown, MA, Williams College, Williamstown. Soil-Survey-Staff (1999) Soil Taxonomy: A Basic System of Soil Classification for Making and Interpreting Soil Surveys USDA Natural Resources Conservation Service. Strange, C.J. (1997) Research Updates: Dark Frontier Beneath Our Feet. BioScience, 47, 339,396. Tillier, S.D.a.P. (1999) Long-term effects of mineral amendments on soil fauna and humus in an acid beech forest floor. Forest Ecology and Management, 118,245-252. Triplehorn, C.A. & Johnson, N.F. (2005) Borror and Delong's Introduction to the Study of Insects, 7 edn. Thomson Learning, Belmont. Trofymow, J.A., Addison, J., Blackwell, B.A., He, F., Preston, C.A., & Marshall, V.G. (2003) Attributes and indicators of old-growth and successional Douglas-fir forests on Vancouver Island. Environmental Reviews, 11, S187-S204. Vitousek, P.M., Mooney, H.A., Lubchenco, J., & Melillo, J.M. (1997) Human Domination of Earth's Ecosystems. Science, 277,494-499. Wall, D.H. & Moore, J.C. (1999) Interactions Underground. BioScience, 49, 109-1 17 Wardle, D.A. (2002) Communities and Ecosystems: Linking the Aboveground and Belowground Components Princeton University Press, Princeton. Weary, G.C. & Merriam, H.G. (1978) Litter Decomposition in a Red Maple Woodlot Under Natural Conditions and Under Insecticide Treatment. Ecology, 59, 180-1 84. Wheater, C.P., Cullen, W.R., & Bell, J.R. (2000) Spider communities as tools in monitoring reclaimed limestone quarry landforms. Landscape Ecology, 15,401-406. Whittaker, R.H. (1970) Communities and ecosystems Macmillan, New York. Wilde, S.A., Voight, G.K., & Iyer, J.G. (1964) Soil and Plant Analysis for Tree Culture Oxford Publ. Hse., Calcutta, India. Willett, T.R. (2001) Spiders and Other Arthropods as Indicators in Old-Growth Versus Logged Redwood Stands. Restoration Ecology, 9,410-420. Williams, P.H. & Gaston, K.J. (1 994) Measuring more of biodiversity: can higher-taxon richness predict wholesale species richness? Biological Conservation, 67,2 1 1-2 17. Wyman, R.L., Posey, E., MacWatters, K.S., Frank, J., & Palange, J. (1999) A new device for extracting living invertebrates from large samples of leaf litter. Applied Soil Ecology, 12,73-79. Yeates, G.W., Hawke, M.F., & Rijkse, W.C. (2000) Changes in soil fauna and soil conditions under Pinus radiata agroforestry regimes during a 25-year tree rotation. Biology and Fertility of Soils, 3 1, 39 1-406. Zen, E. (1983) Bedrock Geologic Map of Massachusetts. APPENDIX Mean plot invertebrate densities and environmental variables 2 1 Burbank 23 Paul Brook Anderson Arbor Acres Bee Fitch Bee Hill Beineke Primary Beineke Rocks Beineke Secondary Bernard Rd Birch Hill Brodie Mid North Brodie North Brodie South Bullock 1933 Bullock WRLF Bullocks Ledge Buxton Hill Caretaker Caves Lot Chenail(22) CL12 Primary CL12 Secondary Deer hdge Secondary East Phelps Field Fann South Flora Glen Ford Glen G113 G114 GI21 G123 G131 GI32 2 1 Burbank 23 Paul Brook Anderson Arbor Acres Bee Fitch Bee Hill Beineke Primary Beineke Rocks Beineke Secondary Bernard Rd Birch Hill Brodie Mid North Brodie North Brodie South Bullock 1933 Bullock WRLF Bullocks Ledge Buxton Hill Caretaker Caves Lot Chenail(22) CL12 Primary CL12 Secondary Deer Rtdge Secondary East Phelps Field Farm South Flora Glen Ford Glen G113 GI 14 G121 G123 G131 G132 + . b:- iT: 21 Burbank 23 Paul Brook Anderson Arbor Acres Bee Fitch Bee Hill Beineke Primary Beineke Rocks Beineke Secondary Bernard Rd Birch Hill Brodie Mid North Brodie North Brodie South Bullock 1933 Bullock WRLF Bullocks Ledge Buxton Hill Caretaker Caves Lot Chenail(22) CL12 Primary CL12 Secondary Deer Ridge Secondary East Phelps Field Farm South Flora Glen Ford Glen GI13 G114 G121 G123 G131 G132 0 C .- V] 2 1 Burbank 23 Paul Brook Anderson Arbor Acres Bee Fitch Bee Hill Beineke Primary Beineke Rocks Beineke Secondary Bernard Rd Birch Hill Brodie Mid North Brodie North Brodie South Bullock 1933 Bullock WRLF Bullocks Ledge Buxton Hill Caretaker Caves Lot Chenail(22) CL12 Primary CL12 Secondary Deer Ridge Secondary East Phelps Field Farm South Flora Glen Ford Glen GI13 GI 14 G121 G123 GI31 G132 * . Qi- w plus grandidentata ~pulus tremuloidies unus pensylvanica unus serotina ims malus uercus alba uercus prinus uercus mbra hamnus cathartics hododendron roseurr hus typhma sbinia pseudo-acaciz 2 1 Burbank 23 Paul Brook Anderson Arbor Acres Bee Fitch Bee Hill Beineke Primary Beineke Rocks Beineke Secondary Bernard Rd Birch Hill Brodie Mid North Brodie North Brodie South Bullock 1933 Bullock WRLF Bullocks Ledge Buxton Hill Caretaker Caves Lot Chenail(22) CL12 Primary CL12 Secondary Deer Ridge Secondary East Phelps Field Farm South Flora Glen Ford Glen G113 GI 14 G121 GI23 GI31 G132 G214 G224 G233 Galusha Prospect Green River North Haley Hopper Hart River Hart Rock HMF Pasture HMF Ryelot Ide rd Mason Hill Lower MGRHS Misery Mtn Money Brook Oaks Loop PCL Petersburg Mtn North Petersburg Mtn South Pine Cobble Upper PK Poritz Schlesinger Scott Hill Stone Hill Stone Hill North Stone Hill South TI24 T133 T214 T221 T23 1 Taconic South G133 G214 G224 G23 3 Galusha Prospect Green River North Haley Hopper Hart River Hart Rock HMF Pasture HMF Ryelot Ide rd Mason Hill Lower MGRHS Misery Mtn Money Brook Oaks Loop PCL Petersburg Mtn North Petersburg Mtn South Pine Cobble Upper PK Poritz Schlesinger Scott Hill Stone Hill Stone Hill North Stone Hill South T124 TI33 T214 T22 1 T23 1 Taconic South G214 G224 G233 Galusha Prospect Green kver North Haley Hopper Hart River Hart Rock HMF Pasture HMF Ryelot Ide rd Mason Hill Lower MGRHS Misery Mtn Money Brook Oaks Loop PCL Petersburg Mtn North Petersburg Mtn South Pine Cobble Upper PK Poritz Schlesinger Scott Hill Stone Hill Stone Hill North Stone Hill South T124 T133 T214 T22 1 T23 1 Taconic South 210 S 186 S 190 S 330 N 260 W 266 W 266 W 96 E 270 W 172 S 125 E 180 S ridge ridge 86 E 105 E 173 S 280 W 76 E 91 E 270 W 290 W 236 W 336 N 5N 90 E 310 W 325 N 202 S 136 S 170 S 56 E 93 E 121 E GI33 G214 G224 G233 Galusha Prospect Green River North Haley Hopper Hart River Hart Rock HMF Pasture HMF Ryelot Ide rd Mason Hill Lower MGRHS Misery Mtn Money Brook Oaks Loop PCL Petersburg Mtn North Petersburg Mtn South Pine Cobble Upper PK Poritz Schlesinger Scott Hill Stone Hill Stone Hill North Stone Hill South T124 T133 T214 T221 T23 1 Taconic South a, . i_l* rn Site G214 G224 G233 Galusha Prospect Green River North Haley Hopper Hart River Hart Rock HMF Pasture HMF Ryelot Ide rd Mason Hill Lower MGRHS Mise~y Mtn Money Brook Oaks Loop PCL Petersburg Mtn North Petersburg Mtn South Pine Cobble Upper PK Poritz Schlesinger Scott Hill Stone Hill Stone Hill North Stone Hill South T124 T133 T214 T22 1 T23 1 Taconic South The Clark Towne Treadwell Hollow TW White Oaks North White Oaks South Wirebridge wow The Clark Towne Treadwell Hollow TW White Oaks North White Oaks South Wirebridge WOW 0 * .- V] Taconic Trail The Clark Towne Treadwell Hollow TW White Oaks North White Oaks South Wirebridge WOW u .+d - V] Taconic Trail The Clark Towne Treadwell Hollow TW White Oaks North Whte Oaks South Wirebridge WOW OOOOOOOOC OOOOOOOOC OOOOOOOOC OOOOOOOOC OP0000h)OC ite cer rubrum cer saccharum cer spicatum melanchier sp. etula allegheniensis etula lenta etula papyrifera etula populifolia arpinus carolina arya cordiformis arya ovata OOOOOOOOC snqolts snu suaqru ea:, ~~~1811.4 ~hz: .ds aa:,gc ealaup sue181 EUVNI~J!A s~iaureute a%FsnU!rn eue:,uaure snulxm E!Io$!Pw% ~n8~ su~snlede3.1!( egojvaqe snuro: eleluap sauelse; aJ!' e39<+++++ 0133