Notes 2630 Ecology, 83(9), 2002, pp. 2630–2636  2002 by the Ecological Society of America DOES HUMAN SCENT BIAS SEED REMOVAL STUDIES? R. SCOT DUNCAN, 1,5 DANIEL G. WENNY, 2 MARK D. SPRITZER, 3 AND CHRISTOPHER J. WHELAN 4 1 Department of Zoology, 223 Bartram Hall, University of Florida, Gainesville, Florida 32611 USA 2 Illinois Natural History Survey, 3159 Crim Drive, Savanna, Illinois 61074 USA 3 Department of Zoology, Miami University, Oxford, Ohio 45056 USA 4 Illinois Natural History Survey, Midewin National Tallgrass Prairie, 30071 South Route 53, Wilmington, Illinois 60481 USA Abstract. Field estimates of seed removal rates are often determined by monitoring the survival of seeds placed at stations. Such experiments may unintentionally provide seed predators with unnatural olfactory cues. We compared the removal of seeds that had direct contact with human skin (scented) vs. seeds that had no contact with human skin (unscented). At three Florida sites in 1997–1998, four to five species were tested by placing five con- specific seeds at each station; 40–50 stations per species, per treatment, per site were monitored for 30 d. Seed removal was greater for scented than unscented seeds, but most differences were significant only with all species and sites pooled. At two Illinois sites and one Florida site in 1999–2000, one seed of one of two species was placed at each of 400 stations and monitored for 30 d. Scented seeds were removed significantly faster than unscented seeds at Illinois sites (14% scented vs. 5% unscented removed after one day), but not at the Florida site. The effect of scent on removal was pronounced during the first week, then disappeared. Fortunately, these results suggest scent biases are weak and short- lived, and most studies are unaffected. Studies that may be affected are those of short duration. Key words: experimental bias; granivore; olfactory cues; seed predation, seed removal. INTRODUCTION Estimations of ecological processes are vulnerable to biases imposed by experimental design. For exam- ple, Cahill et al. (2001) found herbivory estimates were affected by the degree to which plants were handled during surveys. Field estimates of seed removal rates are often determined experimentally by placing seeds at stations and monitoring their survival. Using seed stations has allowed researchers to test hypotheses dif- ficult to study using naturally dispersed seeds. Despite widespread use of seed stations, few studies discuss how accurately these experiments estimate natural seed removal or animal foraging patterns (Wenny 2002). Ro- dents are abundant seed predators in many systems and have an acute sense of smell (Vander Wall 1995, 1998). If researchers leave unnatural olfactory cues (e.g., hu- man scent) on seeds, rodents may detect experimental seeds more than naturally dispersed seeds. This could lead to over- or underestimation of seed removal rates, depending on rodent response to these unnatural cues. Manuscript received 22 June 2001; revised 21 January 2002; accepted 21 January 2002. 5 E-mail: duncan@zoo.ufl.edu Similar studies of avian nest predation use artificial nests and/or artificial eggs to quantify egg predation (Major and Kendal 1996). Several studies have found higher predation rates on artificial than natural nests (Ortega et al. 1998, Sloan et al. 1998, Wilson et al. 1998, Rangen et al. 2000, Zanette and Jenkins 2000, but see Willebrand and Marcstro ¨m 1988, Yahner and Voytko 1989, Skagen et al. 1999). Similarly, Whelan et al. (1994) found predation at artificial nests moni- tored by researchers wearing commercial deer scent (to mask unnatural odors) or scent-neutralizer was lower than predation at nests visited by researchers wearing perfume (a novel stimulus) or no scent manipulation. Human scent may also influence removal rates at ex- perimental seed stations. We compared seed removal rates between seeds that had direct contact vs. no direct contact with human skin. Experiments were conducted in multiple forests in two geographic areas, using 10 species of vertebrate-dispersed seeds among 19 trials. MATERIALS AND METHODS Initial seed removal experiments (experiment 1) were conducted in 1997–1998 at three forests in or near Gainesville, Florida (29° N, 82° W): Natural Area