Spider Size, Web Location, and Prey Capture in the Colonial Orb Weaver Metabus gravidus Jamie E. Potosek Department of Neurobiology and Behavior, Cornell University ABSTRACT Metabus gravidus spiders live in colonies of five to 70 individuals and construct their webs over moving water. Buskirk 1975 found larger spiders can take over the webs of smaller spiders and suggested that certain sites within the colony may have greater prey capture r ates. Therefore, I hypothesized that larger spiders should occupy these positions in the colony. Colonies of M. gravidus were studied at the Quebrada Mquina stream in Monteverde, Costa Rica. I measured spider size, web size, height from water, distance fr om bank, and prey capture rates for 198 total webs. Twenty four artificial webs were created to determine prey capture and web position without the presence of spiders. Spider size and web area correlate positively r = 0.458, p < 0.0001. Larger spiders are found at locations closer to the water surface r = 0.212, p < 0.0001 and catch more prey than smaller spiders r =0.286, p< 0.0001. Larger spiders also show both a higher overall visitation rate by insects to their webs r = 0.231, p = 0.0032 a nd a higher proportion of hits that are successful r = 0.112, p = 0.0002. This data support the earlier idea that web placement within the colony and differential prey capture should place larger spiders in these areas. Buskirk 1975a, Shannon 1996. Ho wever, there is no significance between spider size and distance of web from stream bank r = 0.017, p = 0.0672. Artificial webs showed no significant difference between upstream /downstream p = 0.8614 and high/low positions over the water p= 0.4218 within each colony, but they did show a significantly greater number of prey captures in the inner quadrants of the webs as compared to the outer quadrants p < 0.0001. These findings support the notion that web position alone does not decide the success of prey capture. It is evident that spider size has tremendous influence on success of prey capture as well. RESUMEN Las araas de la especie M. gravidus viven en colonias de cinco a 70 individuos y construyen sus telas arriba del agua moviendo. Buskirk 1975 encontro que las araas mas grandes pueden tomar posesin de las telas de araas mas pequeas. Sugiero que es posible que algunos sitios adentro de colonias tengan proporciones ms altas de capturar presa. Por eso, yo asum una hiptesis que las araas grandes ocuparan estos sitios. Estudie colonias de M. gravidus a la Quebrada Mquina en Monteverde, Costa Rica. Med el tamao de araa, el tamao de tela, la altura sobre el agua, y la distancia de la orilla para 198 telas. Veinticuatro telas artificiales fueron construidas para determinar las capturas de presa y la posicin de la tela sin la presencia de araas. Tamao de araa y area de tela tienen una correlacin positiva r = 0.458, p< 0.0001. Araas mas grande estn encontrado en lugares mas cerca del agua r = 0.212, p < 0.0001 y cogen mas presa que las araas pequeas r = 0.286, p < 0.0001. Las araas mas grandes exhiben mas visitaciones de insectos a sus telas r = 0.231, p = 0.0032 y una proporcin mas alta de visitaciones prosperos r = 0.112, p = 0.0002. Estos resultados apoyan la idea anterior que la colocacin de la tela dentro de la colonia y las capturas de presa deferenciales deben poner las araas ms grandes en estas reas Buskirk 1975a, Shannon 1996. Sin embargo, no hay una diferencia significativa entre el tamao de araa y la
distancia de la tela desde la orilla r = 0.017, p = 0.0672. Las telas artificiales no mostrar on una diferencia significativa entre ro arriba/ ro abajo p = 0.8614 y alta/baja distancias de la superficie del agua p = 0.4218 dentro de cada colonia, pero mostraron una diferencia significativa entre numeros de capturas de presa en las cuadrantes interiores en comparacin de los exteriores p < 0.0001. Estos resultados soportan la nocin que solamente la posicin de tela no decide el xito de capturas de presa. Es evidente que tamao de araa tiene mucha influencia en xitos de capturas de presa t ambin. INTRODUCTION Metabus gravidus is one of a very few orb weaving spider species that is gregarious, constructing its colonies over streams. The proposed benefits of group living include greater efficiency group defense, increased resource exploitation, communal r a ising of young, and the ability to learn from the foraging methods of others Alcock 1984. For M. gravidus coloniality is probably related to resource exploitation and the fac t that single webs cannot be placed over streams. Large populations of M. gravidus can be found in Monteverde, Costa Rica occupying stream banks, and living in colonies of up to 70 individuals suspended over running water Buskirk, 1975a. The colony utili zes common support lines to connect individual webs and facilitate movement within the colony. Each individual spider builds and maintains its own individual web, rarely feeding from other webs in the colony. Prey capture at webs depends on web size, web placement, and the ability of the web to intercept prey, including factors of web angle and stickiness Kajak 1964 et al. cited in Craig 1989. Prior studies of M. gravidus have shown that certain locations within the colonies tend to trap more prey than o thers. Buskirk 1975a observed that in such colonies, web positioned closer to the water surface, farther from the bank, and over slower currents caught more prey than at other web locations. In addition, larger spiders are expected to capture the largest amounts of prey because they will be able to dominate the best locations within the colony. There is evidence that larger spiders are able to do so by initiating aggressive behaviors towards smaller spiders within the colony. Spiders exhibit behaviors of bouncing, web jerks, chasing, displacement of orbs, and fighting which defend individual feeding areas and space out the webs within a colony Buskirk 1975b. These aggressions often result in displacement of the smaller spider or prey robbery from the orb Shannon 1996 found that aggressive behavior only resulted in prey or web acquisition when the aggression was initiated by a larger spider. Building from these previous studies, certain locations within the colonies are expected to be more favorable for prey capture than others. These locations should be upstream, farthest from stream bank, and at the lowest heights above the water. It is also assumed that larger spiders will occupy the superior sites and therefore, experience the highest rates of prey c apture within the M. gravidus colony.
MATERIALS AND METHODS Site Description The study was conducted in premontane wet forest in Monteverde, Costa Rica: Puntarenas province; at an elevation of 1470m. Metabus gravidus colonies were studied along the Quebrada Mquina stream, which at this altitude, averages four meters in width and one half to one meter in depth. The stream is located in protected primary forest near the Estacin Biolgica Monteverde. Spider and Web Siz e Observations were made from 24 October 2000 through 3 November 2000 of spider size in relation to the variables of web area, distance from the bank, height over the water, stream location, and number of prey catches over a set time period to ascertain wh ether the large spiders in fact are able to place themselves preferentially within the colony and catch more prey. Each web within a colony was measured with tape measure for circumference later transformed to area in data analysis, distance from stream bank, and distance above the water surface. In addition, the length of each spider was measured with a caliper. Prey Capture v Real Webs Five similarly measured colonies were subsequently observed from 1600 to 1800 hours and the numbers of insect hits and misses upon each web in the colony were recorded. Hits were defined as being successful prey captures, misses being when insects flew into the web but were not caught. This time period was chosen because Buskirk 1975 observed diurnal changes in insect nu mbers over the water, with increasing numbers in the late afternoon, peaking at 1800 hours. An hour was spent observing each upstream and downstream location in each colony. v Artificial Webs From 6 November 2000 to 15 November 2000, imitation webs were co nstructed at six sites along the same stream, in suitable locations, but where M. gravidus colonies were not present. Four webs were created at each experimental colony. Four strips of contact paper simulating webs were constructed at each Â€colonyÂ, at hig h and low distances above the stream ranging from .08m to 1.08m in height, two upstream and two downstream Figure 1. Webs of equal area were assembled from clear contact paper and covered with automotive grease. After several days, the webs were revisite d and insect
abundances totaled for each web. Web height, distance from bank, and stream location were measured. Distance from bank was recorded in quadrants of each web. Each web was broken up into four quadrants, two closest to the bank, and two farthest from the edge of the stream. Spider size and web size were thus eliminated from the experimental design. RESULTS Spider and Web Size Factors of web location reveal differences in corresponding spider size in the M. gravidus colonies. Spider size Mean + SD = 2.07 + .99cm and web area Mean + SD =1407.32 + 1130.03cm show a positive correlation with a simple regression in this study r = 0.458, p < 0.0001. As spider size increases, web sizes increase accordingly Figure 2 Therefore, just spider size was used in subsequent comparisons. A multiple regression was run on web placement variables versus spider size, with height of web from the water surface Mean + SD = 20.57 + 16.56cm being the only significant abiotic facto r to correlate with spider size r = 0.212, P< 0.0001. Height of web above water was found to correlate negatively with spider size in a simple regression analysis Figure 3. A further simple regression shows that web distance from the stream bank Mean + SD = 193.889 + 68.956cm shares no significant correlation with spider size r = 0.017, p = 0.0672 for the M. gravidus colonies observed Figure 4. Prey Capture v Real Webs Spider size, as well as web location, was found to affect the trends in prey capture in M. gravidus colonies. Analyzed with simple regression, the number of prey captures versus spider size show a positive correlation, r =0.286, p < 0.0001, with larger spiders catching more prey than smaller spiders, ANOVA Test: Figure 5. The number of total visits by potential prey correlates significantly with spider size as well r = 0.231, p < 0.0001. Webs of larger spiders had more visits than those of smaller spiders ANOVA Test Figure 6. The overall proportion of successful prey capt ures defined as the number of hits/total number of visits correlates significantly with spider size as well r = 0.112, p = 0.0002. Larger spiders experience a higher success rate of prey captures than smaller spiders Figure 7.
Prey captures were fo und to be greater at downstream locations Mean + SD = 1.02 + .758 than at upstream locations Mean + SD = .881 + .739, but the difference was not significant when analyzed with an ANOVA post hoc Â‚ test p = 0.3676. A multiple regression analysis showed that the number of prey captures is positively correlated with spider size r = 0.398, p < 0.0001 and distance from stream bank p = 0.013 as well is negatively correlated with height above the water p = 0.0025. v Artificial Webs Web placement alone was not seen to affect prey captures in the artificial Â€coloniesÂ. Each of the four positions within each colony were compared in an ANOVA post hoc test. Upstream and downstream locations show no significant difference in the amount of prey these webs caug ht p = 0.8614 nor did high and low heights above the water p= 0.4218. There was overall no significant difference between the four quadrants sampled Figure 8 in regards to height and stream location. However, a significant difference was found with a paired t test between the inner and outer quadrants of the artificial webs in terms of prey capture p < 0.0001 with inner quadrants showing higher numbers of prey captured than the outer quadrants. DISCUSSION Patterns of web location and prey capture discovered in previous investigations were both upheld and rejected in this study. It has previously been shown that the most desirable web locations within the M. gravidus colonies tend to occur farthest away from the bank, upstream, and closest to the su rface of the running water Buskirk 1975. A negative correlation of height from water surface versus spider size was expected, and the analysis was in fact significant. Larger spiders may prefer lower sites for several reasons. Kerzicnik 1993 found that while M. gravidus is not directly dependent on the water surface for prey capture, it may aid in camouflaging the web from flying insects and protect the spiders from kleptoparasites which rob prey from webs. Spider size did not significantly correlate w ith distance from stream bank, but prey capture was greater at sites farther away from the edge in both the real and artificial webs. Using traps to capture prey, Buskirk 1975a observed that the upstream end of a colony caught more insects relative to do wnstream. The results of this study contrast these earlier findings in that downstream locations tended to catch more prey than upstream locations at the Quebrada Mquina, but the trend was not significant in either the observed or manipulated webs. A poss ible explanation for these inconsistencies may
be that other factors about the colony site have a larger influence on prey capture, such as sunlight intensity and availability of favorable substrates upon which to attach webs. In the artificial web experim ent, exceptionally low prey captures were noted for one colony that was in a large gap, thereby subject to an exceptionally large amount of light. Favorable substrates were certainly more abundant at some colonies than others. This could account for larger amounts of prey capture at these places than otherwise expected taking into account web location. The observation that more prey were caught at a greater distance from the bank is consistent with BuskirkÂƒs earlier findings. It is possible that it is bene ficial for the spiders to be far away from the bank thus locating themselves in the area through which most potential prey fly. It is also possible that the spiders may simply be organizing themselves away from each other in an avoidance of competition, si nce spider size did not vary with distance from bank. BuskirkÂƒs 1975a study found a trend such that the river was narrower in sections where spider colonies were found than those lacking colonies. This was only upheld in the rainy season. This suggests t hat there is perhaps a point at which increased distance from the bank is no longer desirable, and results may differ at streams of different sizes. Body size plays an integral role in web location and number of collected prey. Shannon 1996 found large spiders to catch significantly more prey on a per hour basis than small and medium sized spiders. The data gathered in this study also show a positive correlation between spider size and the number of prey catches. Positive correlations between spider size and total visits, as well as success rate were seen as well. This pattern is well Â‚ documented in many biological systems; larger spiders are able to dominate the locations that are most resource rich Buskirk 1975, Shannon 1996. Clearly within these exi sting colonies, larger spider size is the dominating factor that leads to higher prey capture among M. gravidus Orb weaving spiders rarely feed on each prey caught within the web, rejecting many of the smaller insects that get caught. Foster 1994 obser ved M. gravidus to feed only upon damselflies, which were caught infrequently in the webs. Therefore, prey capture may not be the primary factor in web placement among M. gravidus In Metepeira incrassate, a species of social spider, individual spiders cho ose their web site location by making trade Â‚ offs between foraging success and predation risk Rayor and Uetz 1990 cited in Shelton 1992. It is thus possible that spiders may actively sacrifice the best colony locations in order to avoid predation. Whil e there was a correlation in stream location vs. prey capture for the artificial webs as expected, the data also showed that prey capture did not vary in relation to height above water or stream location. This again supports the idea that spider size may be the
dominating factor in prey capture. However, it is also possible that, in the experimental design the artificial webs were more effective in catching prey and thus stickines s was able to prevail over all other variables. This study supports BuskirkÂƒs 1975a findings that certain web locations in M. gravidus colonies are favored over others in terms of prey capture. Larger spiders were found in these choice sites, and were o bserved to catch more prey than smaller spiders. However, my data also suggests that spider size has a much larger influence on successful prey catches than is apparent in previous studies. In the absence of spiders on the artificial webs, no differences w ere seen in prey capture for height above water and stream location. In conclusion, it is evident that many different factors, both abiotic and biotic, are influential in the colony distribution of M. gravidus spiders in Monteverde, Costa Rica. Much furth er work remains to be done with these unique spiders. All day observations would help to account for temporal differences in activity pattern between young and adult spiders. More extensive experimentation would be beneficial to control for each variable separately. ACKNOWLEDGEMENTS Many thanks must be extended to all of the people who assisted during my project. Thank you to Mauricio Garcia, Andrew Rodstrom, and Tim Kuhman for general, day to day spider advice and supplies yes, car grease really did wo rk despite my doubtfulness!, as well as statistics help; to Alan Masters for all of his insight and much needed guidance from the beginning; and to all my fellow students here in the Fall 2000 program for providing endless entertainment and means of procr astination when it was most needed. Gracias a La Estacin Biolgica Monteverde for use of the stream and forest y mi familia en Caitas por todo. Finally, to my parents, t hanks for making this whole venture possible through your generous funding and suppor t. LITERATURE CITED Alcock, Jon. 1984. Animal Behavior: An Evolutionary Approach. Third Edition. Sinauer Associates, Inc., Sunderland, Massachusetts. Buskirk, Ruth E. 1975. Coloniality, Activity Patterns, and Feeding in a Tropical Orb Weaving Spider. Ecology 56: 1314 1328. Buskirk, Ruth E. 1975. Aggressive Display and Orb Defence in a Colonial Spidar, Metabus gravidus Animal Behavior 23: 560 567. Craig, Catherine. 1989. Alternative Foraging Modes of Orb Web Weaving Spiders. Biotropica 213: 257 264. Foster, Jane. 1994. The Effect of Moisture on Metabus gravidus Webs. CIEE: Tropical Biology and Conservation Program. Summer 1994. 191 197.
Kerzicnik, Laurie. The Relationship of the Water Surface and Prey Capture with the Colonical Orb Weav ing Spider, Metabus gravidus CIEE: Tropical Biology and Conservation Program. Summer 1993. 154 165. Shannon, D.N. 1996. Variations in Colony Activity Among Metabus gravidus Spiders of Different Sizes. UCEAP: Monteverde Tropical Biology Program. Fall 1996. 186 196. Shelton, Angie. 1992. Web Site Tenacity and Behavior of the Social Spiders Metabus gravidus CIEE: Tropical Biology and Conservation Program. Summer 1992. 82 94.