Scatter-hoarding rodents should space food caches to maximize cache recovery rate (to minimize loss to pilferers) relative to the energetic cost of carrying food items greater distances. Optimization models of cache spacing make two predictions. First, spacing of caches should be greater for food items with greater energy content. Second, the mean distance between caches should increase with food abundance. However, the latter prediction fails to account for the effect of food abundance on the behavior of potential pilferers or on the ability of caching individuals to acquire food by means other than recovering their own caches. When considering these factors, shorter cache distances may be predicted in conditions of higher food abundance. We predicted that seed caching distances would be greater for food items of higher energy content and during lower ambient food abundance and that the effect of seed type on cache distance variation would be lower during higher food abundance. We recorded distances moved for 8636 seeds of five seed types at 15 locations in three forested sites in Pennsylvania, USA, and 29 forest fragments in Indiana, U.S.A., across five different years. Seed production was poor in three years and high in two years. Consistent with previous studies, seeds with greater energy content were moved farther than less profitable food items. Seeds were dispersed less far in seed-rich years than in seed-poor years, contrary to predictions of conventional models. Interactions were important, with seed type effects more evident in seed-poor years. These results suggest that, when food is superabundant, optimal cache distances are more strongly determined by minimizing energy cost of caching than by minimizing pilfering rates and that cache loss rates may be more strongly density-dependent in times of low seed abundance.

Despite increasing evidence of seed limitation in forest ecosystems, data remain sparse on spatial patterns of seed rain at large (Ͼ1 ha) spatial scales. We monitored seed rain (28.5 m 2 ) throughout five northern hardwood forest fragments (27 ha sampled across 14-km 2 area) in southeastern Michigan over two years. Four fragments were nearest neighbors (300-700 m), yet varied in species composition, providing the opportunity to detect landscape-scale seed exchange. Of the 37 species of woody plants present in the seed rain (98 032 mature seeds), only three (Betula papyrifera, Ostrya virginiana, and Ulmus americana) had widespread seed dispersions within all fragments containing resident sources (seed in Ͼ70% of traps in each fragment). Seed dispersions, measured as the percentage of traps within a fragment receiving seed, differed among species using different dispersal vectors with animal-dispersed species arriving in a lower percentage of seed traps than wind-dispersed seeds. At a given source density, seed dispersions increased with decreasing seed mass. Light-seeded, fecund species such as Betula or Tsuga required lower source densities to saturate fragments with seed compared to heavy-seeded species (Acer, Fraxinus, Tilia). Heavy-seeded wind-and animal-dispersed species also displayed the strongest evidence of seed limitation, with seedling presence significantly associated with presence of seed for Carpinus caroliniana, Fagus grandifolia, Prunus avium, and Tilia americana. Of 17 species, landscape-scale seed exchange was detected for only four disturbance-associated species (Acer negundo, Betula papyrifera, Celastrus scandens, Eleaganus umbellata). No exchange was detected for Acer rubrum, Betula alleghaniensis, or Tsuga canadensis, despite broad seed dispersions (Ͼ50%) in fragments with resident sources, suggesting the potential for seed limitation for these species at larger spatial scales. Seed encounter probabilities suggest that potential seed competitors often fail to simultaneously colonize microsites. We suggest that all dominant species in northern hardwood forests can be seed limited at some spatial scale and that results are consistent with ''winning by forfeit'' scenarios of diversity maintenance in forest ecosystems.

Numerous studies have documented declines in plant diversity in response to habitat loss in fragmented landscapes. However, determining the mechanisms that lead to species loss is challenging using solely a correlative approach. Here we link correlative assessments of plant community composition with seed additions for a focal species to test the hypothesis that distributions of forests plants within a fragmented landscape are limited by seed dispersal. Woody plant species richness of fragments declined as fragments (n=26) became more isolated by agricultural fields. We predicted that if these isolation effects were driven by poor dispersal rather than other effects associated with habitat loss, then plants should vary in their response to isolation in relation to their seed size (i.e., stronger effects for plants with larger seeds). As predicted under this dispersal limitation hypothesis, sensitivity of bird-dispersed shrubs to isolation was related to their seed mass, with species with heavy seeds (e.g., Lindera benzoin) exhibiting stronger declines in presence across isolation gradients than species with light seeds. Seed addition experiments were performed for Lindera benzoin in two high isolation forest fragments (nearest neighbor mean distance=803 m) where Lindera was naturally absent, and two low isolation fragments (nearest neighbor mean distance=218 m) with naturally occurring Lindera populations. Seed addition and control plots (n=50 1 m 2 plots per fragment) were monitored for 13 censuses over 3 years. Across all four fragments, seed additions resulted in significant increases in Lindera seedling recruitment with no differences in final seedling establishment among fragments. However, insect herbivory was higher on Lindera seedlings in high isolation compared to low isolation fragments and was negatively correlated with seedling survival over some years. Consistent with prior work, our results confirm that seed dispersal plays a significant role in affecting plant diversity in fragmented landscapes. However, results also suggest the need for a better understanding of how additional processes, such as herbivory, may be altered as habitat is lost and what effects such changes have for forest plants.

No abstract

This chapter briefly reviews some of the key challenges in understanding the impact of seed consumers on plant dispersal and demography, and provides an overview of a case study in oaks (Quercus) from Pennsylvania, Maryland, Virginia and southern Indiana, USA, that serves to illustrate many of these issues.