ABSTRACT

EFFECTS OF LANDSCAPE STRUCTURE ON GENERALIST AND SPECIALIST HERBIVORES

by Bradley John Schroeder

The density responses of insect herbivores to shifts in habitat area may depend on the degree of host-plant specialization and the composition of the matrix. I conducted a study in an experimental field comprised of plots of red clover (Trifolium pratense) that differed in size, fragmentation, and surrounding matrix. Densities of three insect herbivores were measured within the plots during two field seasons. All three species showed positive density-area relationships. The two generalist species, the meadow spittlebug ( spumarius) and the Agallian (Agallia constricta), attained higher densities in red clover surrounded by orchard grass (Dactylis glomerata). The specialist species, the clover leafhopper (Ceratagallia agricola), attained higher densities in red clover surrounded by bare ground. The results of this study suggest that both patch area and matrix habitat strongly affect the densities of insect herbivores, and species responses largely depend upon host-plant specialization and the availability of secondary host plants.

EFFECTS OF LANDSCAPE STRUCTURE ON GENERALIST AND SPECIALIST

INSECT HERBIVORES

A Thesis

Submitted to the

Faculty of Miami University

in partial fulfillment of

the requirements for the degree of

Master of Zoology

by

Bradley John Schroeder

Miami University

Oxford, Ohio

2007

Advisor______Dr. Thomas O. Crist

Reader______Dr. Martin Henry H. Stevens TABLE OF CONTENTS

Acknowledgements iii

Effects of Landscape Structure on Generalist and Specialist Insect Herbivores 1 Introduction 1 Methods 4 Results 7 Discussion 8

Tables and Figures 18

ii ACKNOWLEDGEMENTS I would like to thank my advisor, Dr. Tom Crist, for all of his valuable help and advice throughout the course of this project. I would also like to thank my committee members, Dr. Hank Stevens and Dr. Ann Rypstra for their additional advice and suggestions. I would like to thank Rodney Kolb and the staff at the Ecology Research Center for their valuable time and assistance in the field. I would also like to thank Kyle Haynes, Dave Stasek, Laura Douglas and Sam Evans for their hold both out in the field as well as in the lab. This research was funded by the Miami University Zoology Summer Field Workshop.

iii Introduction Understanding species responses to spatial attributes of habitat within the landscape is fundamental to both ecological theory (With and Crist 1995, Ritchie and Olff 1999) and agroecosystem ecology (Tscharntke et al. 2005). Intensive agricultural practices simplify the landscape into large crop monocultures surrounded by semi-natural and natural habitats, causing increases in the abundance of crop pests and a reduction in native species abundances (Robinson and Sutherland 2002). Changes in landscape structure – the size, composition, and spatial arrangement of habitats – due to agricultural development may result in elevated densities of insect herbivores and the loss of natural enemies. Together, these shifts in herbivore or natural-enemy abundance cause reductions in plant biomass and agricultural production (Tscharntke et al. 2005). Generalist predators and non-crop feeding often prefer or require natural and semi-natural areas to undergo life processes, such as overwintering and oviposition (Landis 2000). Many pest insects, however, may complete their entire life cycle within crop fields, and intensive agriculture creates greater areas of habitat for these pest species. Landscape studies of in agricultural systems have focused on the spillover of generalists, particularly generalist predators, from non-crop into crop areas to understand their roles in the control of pest outbreaks (see Kreuss and Tscharntke 1994, Bianchi et al. 2006, Rand et al. 2006). Few studies have compared the responses of insect herbivores differing in habitat or host-plant specialization to habitat area and spatial arrangement (but see Jonsen and Fahrig 1997). Studies of density-area relationships show an increase in population density in larger habitats, a trend that is particularly strong in insect populations (Connor et al. 2000). The resource-concentration hypothesis, proposed by Root (1973) predicts that specialist herbivores in large, monospecific patches of host plants will attain greater relative densities in these simple environments compared to smaller areas with fewer host plants. One primary mechanism for this pattern is based on insect movement, whereby individuals are more likely to find and remain in large stands of host plants, leading to increased densities in larger areas of habitat (Root 1973). Specialist herbivores with life requirements that can be met within the host plants will remain the longest, while those with broader ranges and life requirements will drift out of the patches more quickly.

1 A number of studies have supported positive density-area relationships (e.g. Bach 1984, Kareiva 1985, Kindvall and Ahlén 1992, Matter 1997), while other studies found no relationship and or even negative relationships between insect herbivore densities and habitat area (Bach 1988, Grez & Gonzalez 1995). Hambäck and Englund (2005) have suggested that the inclusion of additional elements of patch geometry (e.g., edge:area ratio) may enhance the predictive value of the resource concentration hypothesis. For species that are visual foragers, for example, immigrants are expected to use patch diameter to locate patches while emigrants use the patch edge, so that population density increases as perimeter-to-area ratio increases, resulting in negative density-area relationships. Alternatively, for species arriving from the air, area-dependent immigration and perimeter-dependent emigration will result in positive density-area relationships. These predictions were supported by Hambäck et al. (in press), who found that moth species, which search for host plants via olfactory information, generally showed positive density-area relationships, while butterflies, which are visual foragers, often showed negative density-area relationships. A reduction in suitable habitat area usually occurs simultaneously with habitat isolation or fragmentation (Carlson & Hartman 2001, Fuller 2001, Summerville and Crist 2001, Fahrig 2003). Therefore, it is difficult to distinguish the effects of habitat loss and fragmentation from one another, and only recently have experiments manipulated fragmentation and area independently (Fahrig 2003). The separation of a given area into various smaller fragments will result in higher edge:area ratios. Species in areas with more edge per given area of habitat will be more likely to leave the habitat and enter the matrix, leading to higher emigration rates and lower population densities within the individual fragments (Fahrig 2002). Apart from habitat area and fragmentation, the type of vegetation within the matrix surrounding habitat patches can have a significant influence on insect herbivore density. Specialist and generalist herbivores may differ in their response to the matrix, because generalists are capable of using complementary resources not available to specialists (With and Crist 1995, Golden and Crist 1999, Haynes et al. 2007). Whereas monophagous insects may have a highly aggregated spatial distribution because they are distributed among patches of their host plant, generalists will be able to use secondary

2 host plants beyond the habitat patches, thus promoting cross-habitat foraging (Shmida and Wilson 1985, Rand et al. 2006). Generalist herbivores may use both the focal habitat as well as the matrix, and become distributed throughout the landscape as though it were a habitat continuum (Tscharntke et al. 2002). The presence of secondary hosts may also encourage these generalist herbivores to remain in habitat patches longer than those that are surrounded by a non-host matrix, leading to higher densities within these areas. Polyculture cropping systems are ideal for examining the effects of plant diversity and the spatial configuration of land-use practices on insect herbivores. Forage crops are often planted as mixtures, such as red clover (Trifolium pratense) with one or more grasses. It is known by practitioners that red clover attracts large numbers of insect herbivores that eventually lead to its decline. In this study, I examined the roles of habitat area, spatial arrangement, and complementary host plants (i.e. grasses) in determining herbivore densities. The dominant herbivores in my study system were three species of : the Agallia constricta Van Duzee and Ceratagallia agricola Hamilton, and the meadow spittlebug Philaenus spumarius (L.). Whereas C. agricola feeds principally on red clover (Hamilton 1998), P. spumarius and A. constricta are more generalized feeders. A. constricta is known to feed upon many herbaceous plants (Osborn 1928). P. spumarius is known to feed preferentially on legumes such as red clover, but it also feeds on a number of grasses and other plant species (Thompson 1994, Weaver and King 1954) and may also be considered a habitat generalist. The objective of this study was to determine how generalist and specialist insect herbivores responded to variations in landscape structure. To test this, I quantified the densities of these three insect herbivores in plots comprised of red clover habitat that differ independently in area (large vs. small), level of fragmentation (continuous vs. fragmented), and matrix composition (orchard grass vs. bare ground). I predicted that insect herbivores would respond to changes in habitat area and fragmentation via changes in density. Additionally, I expected that host-plant specialists would differ from generalists in their responses to matrix composition. Specifically, I predicted that all three species would have greater densities in plots containing large patches of red clover rather than small patches, as predicted by the resource-concentration hypothesis. Second, I predicted that all three insect species would be found in higher densities in plots

3 containing continuous patches rather than fragmented patches, as fragmentation reduces large areas of habitat into several smaller patches with higher edge:area ratio and higher emigration rates. Third, I predicted that the generalists, A. constricta and P. spumarius, would achieve greater densities in plots where the clover patches were surrounded by a matrix of grass compared to those separated by bare ground, as the orchard grass matrix may be considered a secondary resource for both of these species. Since C. agricola is a clover specialist, in contrast, I predicted that there would be no significant effect of matrix type on the density of this species.

Methods Site description This study was conducted at the Miami University Ecology Research Center in a 2-ha field that was planted in the summer of 2004. The field contained a 6 x 6 array of 36 experimental plots, each 14 x 14 m in area, with treatments applied to plots in a completely randomized design. Within each plot were four patches of red clover (Trifolium pretense). There were eight treatments involving manipulations of the patches of red clover and their surrounding matrix within the plots, and an additional set of “control” plots that were planted with a monoculture of orchard grass (Dactylis glomerata) (Fig. 1). Each clover treatment and the grass control were replicated four times. There were three treatment factors for plots containing clover patches, with two levels of each factor. Each plot contained four clover patches that differed in habitat area (4 or 16 m²), fragmentation (continuous or isolated by 2 m) and the surrounding matrix type (orchard grass or bare ground). Clover and grass were established by sowing the plots with seeds of clover and orchard grass (Dactylis glomerata). Plots were maintained by weeding of the clover patches periodically throughout each summer. Grass-matrix areas were kept free from forbs using Weedone LV4 EC 2, 4-D herbicide. The areas of bare-ground matrix, as well as the 8-m bare-ground strips separating the plots were maintained through periodic application of Roundup WeatherMAX herbicide. Finally, the areas of grass were mowed periodically throughout each summer to minimize variation in the height of the grass matrix.

4 Study species P. spumarius, A. constricta, and C. agricola were chosen for this study based on their abundance in the field and their preference for red clover as a host plant. These three species are closely related (Order , Infraorder Cicadomorpha), and all three are phytophagous with differing levels of habitat and host-plant specificity. The leafhopper Agallia constricta Van Duzee is found across much of eastern (DeLong and Davidson 1931). It is macropterous and capable of flight (Hamilton 1998). It is considered a habitat generalist that feeds upon the phloem of many low herbaceous plants (Osborn 1928). Its peak abundance within the field was in mid- to late July. The clover leafhopper, Ceratagallia agricola Hamilton, reproduces continuously throughout the summer (Hamilton 1998). This species is found across much of North America (DeLong and Davidson 1931). It is also macropterous and adept at flight. It is also a phloem-feeding Hemipteran. The chief host of this insect is red clover and therefore can be considered a host-plant specialist (Hamilton 1998). C. agricola adults reached their peak abundance in early September. The meadow spittlebug Philaenus spumarius (L.) is a univoltine species, which hatches in late spring and early summer (Weaver and King 1954). It has a sedentary nymphal stage within a microhabitat of spittle that it creates by releasing excess water while feeding on xylem at the main transpiration stream of plants (Malone et al. 1999). This spittle mass protects it from desiccation, thermal stress, water loss, attacks by predators and parasitoids, and dislodgement from its host (Whittaker 1970). As an adult, the spittlebug has the same feeding habits, but no spittle collects around the body (Wiegert 1964). Although it is considered to be a generalist and feeds on a wide variety of plants, it is also known to have a strong preference for legumes such as red clover (Thompson 1994). Because it uses the orchard grass as a host plant, however, I considered the meadow spittlebug a habitat generalist in this study. This species reached its peak abundance in early June. Density measurements Density measurements of the three insect species were obtained in summer 2005 and 2006. Samples of the insects were taken within each replicate plot during the period of their respective peak abundance (early June for P. spumarius, mid-July for A.

5 constricta, and early September for C. agricola). The insects were collected with a D- Vac® insect vacuum. Within each clover patch, the circular 0.08 m² sampling hose was placed over the vegetation at four stratified random locations. This sampling technique was repeated within the area of matrix surrounding each of the patches, for a total of 32 suctions per plot. During the last two sample periods, in mid-July and early September of 2006, other experiments conducted within the plots prevented the use of two patches within each plot for density sampling. For this reason, only two patches and their surrounding matrix were sampled during these two sampling periods, for a total of 16 suctions per plot. During the 2006 growing season, there was high mortality of clover in several experimental plots. One plot in July and an additional ten plots in September had high clover mortality, and no insects were collected from these plots. Data Analysis I evaluated the effects of clover habitat area, fragmentation, and matrix composition on insect densities in clover using a factorial three-way ANOVA. Prior to analysis, density of the insects throughout the entire plot were expressed by finding the average density of the suction samples from the red clover, and then using this density value for the 0.08 m² sampling area to determine the abundance of insects within the entire area of clover within the plot. In the same manner, the average density of suction samples from the matrix was used to determine total abundance of insects within the matrix of the plot. The abundance of insects within the clover and the matrix were combined, and then divided by the area of the entire plot, in order to determine the density of the insects at the level of the plot. Plot-level densities were ln-transformed to improve homogeneity of variances among treatments and normality of the distribution of the error. Densities within clover patches were also compared against the matrix habitats via paired Wilcoxon tests (Hollander & Wolfe, 1973), because densities in clover and the adjacent matrix in the same plot were not considered to be statistically independent. Finally, grass-matrix densities were compared with bare-ground densities to determine which matrix type was preferred for each of the three species. Bare-ground and grass- matrix densities were compared using unpaired Wilcoxon tests. Each of these statistical comparisons was conducted separately for the three insect species and two sample years.

6 Results Overall, sampling from 2005 and 2006 yielded a total of 361 individuals of P. spumarius, 7,599 of A. constricta, and 738 of C. agricola. In clover patches, average densities of P. spumarius were 4.3 ± 4.3 /m² (± 1 SD) in 2005 and 2.5 ± 2.3 /m² in 2006. Densities of A. constricta were 135.1 ± 64.3 /m² in 2005 and 60.6 ± 29.2 /m² in 2006. Finally, the densities of C. agricola were 15.1 ± 7.4 /m² in 2005 and 7.3 ± 6.1 /m² in 2006. P. spumarius P. spumarius density was significantly influenced by clover patch area in 2005 (P = 0.001, Table 1). Densities in plots containing large patches were 3.8 times higher than in plots with small patches (Fig. 2). In 2006, P. spumarius responded to matrix type (P < 0.001), with densities 8.8 times higher in plots with a grass matrix than in those with bare ground. There was also a significant interaction of matrix type with habitat area (P = 0.001). P. spumarius did not show any distinct difference in density between large and small patches of red clover when the patches were surrounded by a grass matrix. Plots with a bare-ground matrix, however, had higher densities in plots with large patches as opposed to small patches. A. constricta In 2005 A. constricta showed significant responses to both patch area (P < 0.001, Table 2) and matrix type (P < 0.001). A. constricta densities were 1.8 times higher in plots containing large patches of clover than in plots with small patches (Fig. 3, P < 0.001). A. constricta densities were 2.4 times higher in plots with a grass matrix than in those with bare ground. There was also an interactive effect between area of clover patches and matrix type (P = 0.029). Densities of A. constricta were higher in plots with large patches of clover, when the plots contained a bare ground matrix (Fig. 3). In plots with a grass matrix, however, there was no effect of area. A. constricta showed significant responses to patch area (P < 0.001) and matrix type (P < 0.001) again in 2006. Densities of A. constricta were 2.5 times higher in plots with large patches of red clover than in plots containing small patches. Similarly, densities in plots with a grass matrix were 2.5 times higher than in plots with a bare-ground matrix.

7 C. agricola In 2005, C. agricola densities were strongly affected by patch area (P < 0.001, Table 3), and matrix type (P < 0.001). C. agricola densities were 1.7 times higher in plots with large clover patches than those with small patches (Fig. 4). C. agricola densities were also 1.8 times higher in plots with a bare-ground matrix than in those containing a grass matrix. Additionally, there was a significant area by fragmentation interaction (P = 0.013). C. agricola densities responded to patch area when the patches were continuous, but did not show a significant response when the patches were fragmented. C. agricola densities showed significant responses to patch size (P = 0.009) and matrix type (P < 0.001) again in 2006. Densities of C. agricola were higher in plots with large patches than in those with small patches. Densities were also higher in plots with a bare ground matrix compared to those with a grass matrix. Comparisons of clover to surrounding matrix The densities of all three species were significantly higher in clover than in the matrix for both grass and bare-ground matrix types for both years (Table 4). Comparisons of insect abundances between the two matrix types showed no significant difference in P. spumarius densities between grass and bare ground in 2005 (Table 5) (Fig. 5). In 2006, however, P. spumarius densities were 50 times higher in grass matrix than in the bare ground. A. constricta densities were 23.5 and 14 times higher in grass than in bare ground during 2005 and 2006, respectively (Fig. 6). In contrast to the two generalist herbivores, however, densities of the specialist C. agricola were 3.7 times higher in the bare-ground matrix than in grass during 2005 and 9.1 times higher during 2006 (Fig. 7).

Discussion Responses to area Each of the three species examined in this study showed differences in density in response to habitat area during one or both field seasons. As predicted, an increase in habitat area resulted in higher herbivore density within the plots. These positive density- area relationships are consistent with Root’s (1973) resource-concentration hypothesis, suggesting that these three insect herbivores were more likely to find and remain in large

8 stands of host plants than small ones. The response to patch area did, however, vary from year to year and with the inclusion of additional elements such as matrix type and fragmentation. Hambäck and Englund (2005) incorporate features such as patch perimeter and edge:area ratios into models that predict density-area relationships. These models may help to explain variation among the responses to habitat area by examining the movement of the insect herbivores in response to fragmentation and matrix type. Alternative explanations for the positive density-area relationship have also been proposed, such as the enemies hypothesis (Root 1973), which suggests that predation risk may be higher in small patches, leading to lower herbivore densities in those patches. To test the enemies hypothesis, I compared densities of Agallia constricta and those of damsel bugs (Nabis spp.), which was a common generalist predator collected in these samples. However, I found no relationship between leafhopper and damsel-bug densities across treatments. The densities of spiders collected within the D-vac samples in July of 2005 were also examined in response to the treatment types. Overall density for all spider guilds (running, ambush, and web-building) was highest in large patches of clover (J. Schmidt, unpublished data). However, if predator densities in the small clover patches were initially high, than the resulting predator pressure may have resulted in lower prey densities, leading to a decrease in predator densities over time due to lack of available prey. It is unclear, therefore whether these results support or refute the enemies hypothesis. It may be necessary to monitor predator densities throughout the field season in order to determine if the low herbivore densities in plots with small clover patches is a result of an initially high predation risk. Connor et al. (2000) give a number of alternative suggestions for positive density- area relationships beyond dispersal and predation. Elements such as habitat quality and habitat specialization have been known to affect population densities. In the present study, however, I held host-plant composition and plant density constant. Nonetheless, more subtle variations in host-plant quality, such as differences in nutrient availability, may be important within host plant species (Elser et al. 2000). Other factors such as higher probabilities of over-winter survival or annual re-colonization may also influence the densities of insect herbivores, and may vary in response to changes in habitat size (Raupp & Denno 1979). Studies that more thoroughly test alternative mechanisms and

9 possible advantages of larger habitat patches are needed to fully explain the positive density-area relationships that occur. Responses to matrix type The type of matrix surrounding the focal habitat patches had a significant effect on the densities of the three herbivores. The use of complementary host resources would explain the response of the insects to the matrix. P. spumarius and A. constricta are generalist herbivores that use a broader range of host-plant species. Comparisons of the densities of each species within the grass and bare-ground matrix types indicate that both P. spumarius (Figs. 8 & 9) and A. constricta (Figs. 10 & 11) occur in higher densities in the grass matrix than in the bare ground. The presence of secondary host plants within the matrix allows for the generalist species to access varying types of habitat containing required resources. Greater access to different important resources may explain why these species had higher overall densities in plots with a matrix of secondary vegetation compared with those that lack vegetation. These “spillover” effects between host-plant patches that differ in suitability are now increasingly recognized as important determinants of insect herbivores in agricultural landscapes (Rand et al. 2006). In contrast, the specialist leafhopper, C. agricola, was found in higher densities in patches that were surrounded by bare-ground matrix. Competition of red clover with companion plants may reduce biomass of clover in patches surrounded by a grass matrix (Hooks & Johnson 2003). This reduction in the biomass of the host plants may cause the densities of specialist species to be lower than in patches of host plants that are surrounded by bare ground. The bare-ground matrix may also have an effect on nutrient quality of the host plants. Haynes and Cronin (2003) found that patches of cordgrass supported higher densities of delphacid planthoppers (Prokelisia crocea) when the patches were surrounded by a mudflat matrix rather than that of non-host grasses. The mudflat –embedded cordgrass patches also had higher leaf-nitrogen content than those surrounded by non-host grasses, suggesting that leafhoppers were more likely to remain in patches with a mudflat-matrix due to higher availability of nutrients. Insects that feed on red clover, however, are not expected to respond to leaf-nitrogen content, as red clover is a nitrogen-fixing plant and is expected to have higher than average nitrogen content when compared to other terrestrial plants. Samples of red clover from the experimental

10 field were found to have an average C:N ratio of 10.22 ± 0.97, which is considerably lower and with lower variability than the average C:N ratio of 36 ± 23 among terrestrial autotrophs (Elser et al. 2000). Therefore it seems unlikely that insects feeding primarily on red clover will be nitrogen limited or respond to changes in leaf-nitrogen content. Comparisons between the grass and bare-ground densities provide evidence that C. agricola makes occasional use of the bare ground, more so than the grass matrix (Figs. 12 & 13). In contrast to the grass matrix, the bare ground may facilitate a necessary function for C. agricola, such as thermoregulation (May 1979), encouraging greater movement into the bare ground matrix as well as supporting higher overall densities of the herbivores by providing a more favorable thermal microenvironment. Responses to Fragmentation The specialist species, C. agricola, showed an interactive response of habitat fragmentation and habitat area during the first field season. The clover leafhopper responded to patch area with an increase in density, but only among continuous patches. Densities in fragmented patches remained low among plots of both large and small patches. Fragmentation of the clover patches created four small, distinctly separated patches with greater edge:area ratios rather than one large, continuous area of habitat. These smaller individual areas with greater edge:area ratios may have encouraged greater emigration rates and lower overall densities of insect herbivores (Tscharntke et al. 2002, Fahrig 2002). In plots where the large patches were reduced to several smaller fragments, the insects responded to these as though they were each a small habitat patch, and had lower densities within this fragmented habitat. These results were not consistent between the two field seasons. However, the loss of several replicates may have made it more difficult to detect interactions between the two treatments. Neither generalist species studied displayed any responses to the fragmentation of habitat among the two field seasons. A review by Debinski and Holt (2000) has shown a considerable lack of consistency in results provided by fragmentation experiments. Possible explanations for this include crowding effects in fragmented patches. Additionally, for generalist species, increased exposure to the matrix via higher edge:area ratios might result in greater access to complementary resources within the matrix. Many of the experiments reviewed by Debinski and Holt (2000) did not vary area and

11 fragmentation independently, however, so that several previous studies may actually be showing area rather than fragmentation effects (Fahrig 2003). Mark-release recapture data on the three insect species suggested that individuals were transient within experimental patches, whole plots, and perhaps even the entire field. More than 90% of the leafhoppers released into a patch within the field had left these source patches in the first 12 h, as well as almost 60% of the spittlebugs released. These data suggest that all three species used several experimental patches with frequent foraging movements among plots within the 2-ha field and the broader landscape. At broader spatial scales, the type and amount of resources available would affect insect densities to a greater degree than the spatial distribution of resources within the plots (Summerville and Crist 2001). This may explain why patch size, but not fragmentation, had significant effects on the densities of the two generalist species. Responses of herbivores to variation in landscape structure The responses of the generalist and specialist herbivores to patch area suggest that the presence of smaller areas of cultivated crops will result in decreased densities of both generalist and specialist insect herbivores. Both generalist and specialist herbivores showed strong responses to an increase in habitat area with an increase in density. Additionally, our results demonstrate that matrix type around crop monocultures will have a significant impact on herbivore densities within these crop areas. The generalist species in this study responded to adjacent complementary habitat with an overall increase in density. In contrast, specialist species reached higher densities within monocultures surrounded by a matrix lacking additional plants. The results of this study suggest that while maintaining crop monocultures does encourage outbreaks of specialist herbivores, the addition of alternative host plants via agricultural practices such as mixed forage crops or intercropping may promote outbreaks of generalists as it lowers the densities of specialist species. Enhanced control of both specialist and generalist insect herbivores in agroecosystems, especially highly vagile insects that respond to the type and amount of habitat available on a broad scale, may require careful dilutions of primary host plants with secondary plant species in order to minimize the densities of both generalist and specialist within crop fields. Additionally, diversified landscapes that support alternative hosts and prey may also support an

12 increase in predator populations, leading to improved pest control (Settle et al. 1996, Bianchi and van der Werf 2004, Östman 2004, Bianchi et al. 2006). Maintaining smaller patches of cropland rather than expansive crop monocultures may also assist in lowering the densities of both generalist and specialist herbivore species. The role of fragmentation in controlling herbivore densities may vary significantly depending on the vagility of the species. While highly transient generalist herbivores may be unaffected by fragmentation on smaller scales, more dispersal-limited species and those that specialize on a particular crop may be impeded by separation of agricultural land and the presence of intervening non-habitat. Future examination of species response to fragmentation will require a multi-scale approach, one that can examine the responses of herbivores with differing dispersal modes to changes in landscape fragmentation at multiple scales. The results of this study demonstrate that insect herbivores respond to multiple elements of landscape structure, and that these responses may vary depending upon their degree of specialization.

13 References

Bach, C. E. (1984) Plant spatial pattern and herbivore population dynamics: plant factors affecting the movement patterns of a tropical cucurbit specialist (Acalymma innubum). Ecology, 65, 175-190.

Bach, C. E. (1988) Effect of host plant patch size on herbivore density: patterns. Ecology, 69, 1090-1102.

Bach, C. E. (1988) Effects of host plant patch size on herbivore density: underlying mechanisms. Ecology, 69, 1103-1117.

Bianchi, F. J. J. A., Booij, C. J. H., and Tscharntke, T. (2006) Sustainable pest regulation in agricultural landscapes: a review on landscape composition, biodiversity, and natural pest control. Proceedings of the Royal Society B, 273, 1715-1727.

Bianchi, F. J. J. A. and van der Werf, W. (2003) Model evaluation of the function of prey in non-crop habitats for biological control by ladybeetles in agricultural landscapes. Ecological Modelling, 171, 177-193.

Carlson, A. and Hartman, G. (2001) Tropical forest fragmentation and nest predation- an experimental study in an Eastern Arc montane forest, Tanzania. Biodiversity Conservation, 10, 1077-1085.

Connor, E. F., Courtney, A. C. and Yoder, J. M. (2000) Individuals-area relationships: the relationship between population density and area. Ecology, 81, 734- 748.

Debinski, D. M. and Holt, R. (2000) A survey and overview of habitat fragmentation experiments. Conservation Biology, 14, 342-355.

DeLong, D. M. and Davidson, R. H. (1931) The genus Agallia- External characters used to distinguish the species injuring economic crops. Ohio Journal of Science, 31, 377-384.

Elser, J. J., Williams, F. F., Denno, R. F., Dobberfuhl, D. R., Folarin, A., Hubert, A., Interland, S., Kilham, S. S., McCauley, E., Schulz, K. L., Siemann, E. H., and Sterner, R. W. (2000) Nutritional constraints in terrestrial food webs. Nature, 408, 578-580.

Fahrig, L. (2002) Effect of habitat fragmentation on the extinction of threshold: a synthesis. Ecological Applications, 12, 346-353,

Fahrig, L. (2003). Effects of habitat fragmentation on biodiversity. Annual Review of Ecology, Evolution, and Systematics, 34, 487-515.

14

Fuller, D. O. (2001) Forest fragmentation in Loundoun County, Virginia, USA evaluated with multitemporal Landsat imagery. Landscape Ecology, 16, 627-642.

Golden, D. M. and Crist, T. O. (1999) Experimental effects of habitat fragmentation on old-field canopy insects: community, guild, and species responses. Oecologia, 118, 371 380.

Grez, A. A. and Gonzalez, R. H. (1995) Resource-concentration hypothesis: effect of host plant patch size on density of herbivorous insects. Oecologia, 103, 471-474.

Hambäck, P. A. and Englund, C. (2005) Patch area, population density and the scaling of migration rates: the resource concentration hypothesis revisited. Ecology Letters, 8, 1057-1065

Hambäck, P. A., Summerville, K. S., Steffan-Dewenter, I., Krauss, J. Englund, G., and Crist, T. O. (2007) Habitat specialization, body-size and family identity explain Lepidopteran density-area relationships in a cross-contintental comparison. Prcoeedings of the National Academy of Sciences, in press.

Hamilton, K. G. A. (1998) The species of the North American leafhoppers Ceratagallia kirkaldy and Aceratagallia kirkaldy (Rhynchota: Homoptera: Cicadellidae). The Canadian Entomologist, 130, 427-490.

Haynes, K. J. and Cronin, J. T. (2003) Matrix composition affects the spatial ecology of a prarie planthopper. Ecology, 84, 2856-2866.

Haynes, K. J., Diekötter, T., and Crist, T. O. (2007) Resource complementation and the response of an insect herbivore to habitat area and fragmentation. Oecologia, 153, 511-520

Hollander, M. and Wolfe, D. A. (1973) Nonparametric Statistical Methods. Wiley, New York.

Hooks, C. R. R., and Johnson, M. W. (2003) Impact of agricultural diversification on the insect community of cruciferous crops. Crop Protection, 22, 223-238.

Kareiva, P. M. (1985) Finding and losing host plants by Phyllotreta: patch size and surrounding habitat. Ecology, 66, 1809-1816.

Kindvall, O. and Ahlen, I. (1992) Geometrical factors and metapopulation dynamics of the bush cricket, Metrioptera bicolor philippi (Orthoptera: Tettigoniidae). Conservation Biology, 4, 520-529.

Kreuss, A. and Tscharntke, T. (1994) Habitat fragmentation, species loss, and biological control. Science, 264, 1581-1584.

15

Jonsen, I. D. and Fahrig, L. (1997) Response of generalist and specialist herbivores to landscape spatial structure. Landscape Ecology, 12, 185-197.

Landis, D. A., Wratten, S. D., and Gurr, G. M. (2000) Habitat management to conserve natural enemies of pests in agriculture. Annual Review of Entomology, 45, 175-201.

Malone, M., Watson, R., and Pritchard, J. (1999) The spittlebug Philaenus spumarius feeds from mature xylem at the full hydraulic tension of the transpiration stream. New Phytology, 143, 261-271.

Matter, S. F. (1997) Population density and area: the role of between- and within-patch processes. Oecologia, 110, 533-538

May, M. L. (1979) Insect thermoregulation. Annual Review of Entomology, 24, 313- 349.

Osborn, H. (1928) The leafhoppers of Ohio. Ohio Biological Survey. Bulletin 14.

Östman, Ö. (2004) The relative effects of natural enemy abundance and alternative prey abundance on aphid predation rates. Biological Control, 30, 281-287.

Rand, T. A., Tylianakis, J. M., and Tscharntke, T. (2006) Spillover edge effects: the dispersal of agriculturally subsized insect natural enemies into adjacent natural habitats. Ecology Letters, 9, 603-614.

Raupp, M. J. and Denno, R. F. (1979) The influence of patch size on a guild of sap feeding insects that inhabit salt marsh grass Spartina pratens. Environmental Entomology, 8, 412-417.

Ritchie, M. E. and Olff, H. (1999) Spatial scaling laws yield a synthetic theory of biodiversity. Nature, 400, 557-560.

Robinson, R. A. and Sutherland, W. J. (2002) Post war changes in arable farming and biodiversity in Great Britain. Journal of Applied Ecology, 39, 157-176.

Root, R. B. (1973) Organization of a plant-arthropod association in simple and diverse habitats: the fauna of collards (Brassica oleracea). Ecological Monographs, 43, 95-124.

Settle, W. H., Ariawan, H., Astuti, E. T. Cahyana, W., Hakim, L. A., Hindayana, D., Lestari, A. S., and Pajarningsih. (1996) Managing tropical rice pests through conservation of generalist natural enemies and alternative prey. Ecology, 77, 1975-1988.

16 Shmida, A., and Wilson, M. V. (1985) Biological determinants of species diversity. Journal of Biogeography, 12, 1-20.

Simberloff D. (1988) The contribution of population and community biology to conservation science. Annual Review of Ecology and Systematics, 19, 473-511.

Summerville, K. S. and Crist, T. O. (2001) Effects of experimental habitat fragmentation on patch use by butterflies and skippers (Lepidoptera). Ecology, 82, 1360-1370.

Thompson, V. (1994) Spittlebug indicators of nitrogen-fixing plants. Ecological Entomology, 19, 391-398.

Tscharntke, T, Steffan-DeWenter, I., Kruess, A., and Thies, C. (2002) Characteristics of insect populations on habitat fragments: a mini-review. Ecological Research, 17, 229-239.

Tscharntke, T., Klein, A. M., Kruess, A, Steffan-DeWenter, I. and Thies, C. (2005) Landscape perspectives on agricultural intensification and biodiversity – ecosystem service management. Ecology Letters, 8, 857-874.

Weaver, C. R. and King, D. R. (1954) Meadow spittlebug. Ohio Agricultural Experiment Station, Wooster Research Bulletin. 741.

Whittaker, J. B. (1970) Cercopid spittle as microhabitat. Oikos, 21, 59-64.

Wiegert, R. G. (1964) The ingestion of xylem sap by meadow spittlebugs, Philaenus spumarius (L.). American Midland Naturalist, 71, 422-428.

With, K. A. and Crist, T. O. (1995) Critical thresholds in species’ responses to landscape structure. Ecology, 76, 2446-2459.

17 Table 1. Results of three-way ANOVA of ln-transformed densities of P. spumarius in 2005 and 2006.

2005: 2006:

Treatment df MS F P-value df MS F P-value Area 1 25.851 13.83 < 0.001 1 0.764 2.34 0.139

Matrix 1 0.382 0.33 0.569 1 41.459 127.24 <0.001

Fragmentation 1 0.451 0.39 0.537 1 0.616 1.89 0.182

Area*Matrix 1 0.134 0.12 0.735 1 4.379 13.44 < 0.001

Area*Fragmentation 1 0.982 0.86 0.364 1 1.125 3.45 0.075

Matrix*Fragmentation 1 0.233 0.20 0.656 1 1.103 3.39 0.078

Area*Matrix*Fragmentation 1 0.477 0.42 0.525 1 0.151 0.46 0.503

Error 24 1.146 24 0.3258

18 Table 2. Results of three-way ANOVA of ln-transformed densities of A. constricta in 2005 and 2006.

2005: 2006:

Treatment df MS F P-value df (MS) F P-value Area 1 4.072 25.01 < 0.001 1 7.191 51.19 < 0.001

Matrix 1 7.616 46.78 < 0.001 1 7.462 53.11 < 0.001

Fragmentation 1 0.143 0.88 0.358 1 0.061 0.43 0.516

Area*Matrix 1 0.883 5.42 0.029 1 0.336 2.39 0.136

Area*Fragmentation 1 0.004 0.03 0.872 1 0.034 0.25 0.625

Matrix*Fragmentation 1 0.026 0.16 0.692 1 0.057 0.40 0.531

Area*Matrix*Fragmentation 1 0.007 0.04 0.839 1 0.195 1.39 0.250

Error 24 0.1628 23 0.1405

19 Table 3. Results of three-way ANOVA of ln-transformed densities of C. agricola in 2005 and 2006.

2005: 2006:

Treatment df MS F P-value df MS F P-value Area 1 2.941 13.46 0.001 1 5.669 9.36 0.009

Matrix 1 3.844 17.59 <0.001 1 14.471 23.88 <0.001

Fragmentation 1 0.350 1.60 0.218 1 0.000 0.00 0.980

Area*Matrix 1 0.407 1.86 0.185 1 1.065 1.76 0.208

Area*Fragmentation 1 1.576 7.21 0.013 1 0.044 0.07 0.792

Matrix*Fragmentation 1 0.017 0.08 0.782 1 0.098 0.16 0.695

Area*Matrix*Fragmentation 1 0.147 0.67 0.420 1 0.952 1.57 0.232

Error 24 0.2185 13 0.6059

20 Table 4. Results of paired Wilcoxon test comparing the densities of three insect herbivores within clover patches to the densities of the insects within the matrix surrounding the patches.

Species Year W* Clover/Grass W* Clover/Bare Ground P. spumarius 2005 -3.08 P = 0.0010 -4.753 P < 0.0001

2006 -8.401 P = 0.0344 -4.853 P < 0.0001

A. constricta 2005 -4.93 P < 0.0001 -4.84 P < 0.0001

2006 -1.802 P < 0.0001 -3.874 P < 0.0001

C. agrícola 2005 -4.84 P < 0.0001 -4.87 P < 0.0001

2006 -3.435 P = 0.0003 -4.00 P < 0.0001

21 Table 5. Results of unpaired Wilcoxon comparisons of the densities of three insect herbivores within bare ground and grass matrix types.

Species sampled within matrix W P-value P. spumarius, 2005 230.0 P = 0.0950

P. spumarius, 2006 138.5 P < 0.0001

A. constricta, 2005 136.0 P < 0.0001

A. constricta, 2006 150.5 P = 0.0001

C. agricola, 2005 334.5 P = 0.0048

C. agricola, 2006 152.0 P = 0.0130

22

Fig. 1. Diagram of the experimental model landscape.

23

Fig. 2. The effects of habitat area, fragmentation, and surrounding matrix type on the density of P. spumarius in June of 2005 and June of 2006. Data are means ± 1 S.E.

24

Fig. 3. The effects of habitat area, fragmentation, and surrounding matrix type on the density of A. constricta in July of 2005 and July of 2006. Data are means ± 1 S.E.

25

Fig. 4. The effects of habitat area, fragmentation, and surrounding matrix type on the density C. agricola in September of 2005 and September of 2006. Data are means ± 1 S.E.

26

Fig. 5. Densities of P. spumarius in June of 2005 and June of 2006 within the clover patches and the surrounding matrix. Densities within the control grass plots are shown for comparison. Data are means + 1 S.E.

27

Fig. 6. Densities of A. constricta in July of 2005 and July of 2006 within the clover patches and the surrounding matrix. Densities within the control grass plots are shown for comparison. Data are means + 1 S.E.

28

Fig. 7. Densities of C. agricola in September of 2005 and September of 2006 within the clover patches and the surrounding matrix. Densities within the control grass plots are shown for comparison. Data are means + 1 S.E.

29