Seed Removal Rates Under Isolated Trees and Continuous Vegetation in Semiarid Thornscrub

Abstract Clearing native vegetation to increase the amount ofland available for agriculture in northeastern Mexicohas left remnants ranging in size from fragments of con-tinuous vegetation to isolated individual trees. Theseprovide valuable opportunities for restoring larger areasof native vegetation. We explored whether fragmenta-tion of Tamaulipan thornscrub affects the removal ofseeds from 12 woody species that encompass a range ofsizes and dispersal mechanisms. We tested whether (1)seed removal rates under isolated trees were higher thanunder continuous vegetation; (2) dispersal structures,such as fleshy pericarps, made some seeds more attrac-tive to seed removers; and (3) microenvironmental vari-ation affected seed removal rates. Seeds were placedunder canopies of Texas ebony (Ebenopsis ebano)andMesquite trees (Prosopis laevigata). Seed removal trialswere conducted three times, each trial lasting 30 days.Most seeds were removed in all trials by the end of onemonth. Seed removal rate was slower under isolatedtrees. In general, fleshy fruits were removed faster thanother fruits; whole fruits and fleshy tissue were removedfaster than depulped seeds. In species with fleshy peri-carps, acid washing of seeds, to simulate seeds processedin the digestive tract of dispersers, reduced the seedremoval rates, suggesting that it would be a good pre-treatment for restoration efforts.Key words: depulped seeds, fleshy pericarps, fragmentation,restoration, seed dispersal, Tamaulipan thornscrub.IntroductionThe pace and magnitude of landscape transformation inrecent decades has resulted in increasingly fragmentednatural ecosystems. This fragmentation of habitats isexpected to have significant impacts on biodiversitybecause remnant habitat patches become smaller andmore isolated (Burgess & Sharpe 1981; Rudd et al. 2002).Several studies in the past decade have dealt with theimplications of such losses; some focused on a theoreticalperspective (Forman 1995; Henle et al. 2004a; Melbourneet al. 2004), whereas others examined the effect of habitatfragmentation on species distribution and abundance(Hansen & Di Castri 1992; Hobbs & Saunders 1993;Laurance & Bierregaard 1997; Schwartz 1997; Henle et al.2004b and references therein). The major findings suggestthat (1) smaller fragments tend to have fewer species andsmaller populations than larger fragments or continuousvegetation (Jacquemyn et al. 2002, 2003; Hill & Curran2003); (2) species in smaller fragments have a higher riskof extinction because of environmental, genetic, anddemographic stochasticity (Pimm et al. 1988; Menges1991; Young & Clarke 2000); (3) some species are favoredby smaller fragments, whereas others only remain in largefragments (Kelly et al. 2000; Fischer & Lindenmayer2002); (4) fragmentation induces changes in vital speciesinteractions because, for instance, fewer pollinators havebeen found in smaller fragments (Aizen & Feinsinger1994a, 1994b); and (5) the (re)colonization potential ofsmaller fragments is reduced either by the surroundingunsuitable habitat or by isolation from propagule sources(Primack & Miao 1992; Gorchov et al. 1993; McClanahan& Wolfe 1993; Jacquemyn et al. 2003). Collectively, theevidence indicates that small habitat patches are likely tolose some species and become impoverished.Habitat restoration has been examined for its potentialto stop fragmentation and the resulting loss of species.Research has focused particularly on the restoration offorest fragments in temperate and tropical ecosystems(e.g., Peterken 2000; Lusk & Kelly 2003); less research hasexamined restoration, conservation, and reforestation offragmented landscapes in semiarid or arid regions wherehigh abiotic stress contributes to severe recruitment anddispersal limitations and high mortality of individualplants. The most critical factor controlling plant growthand reproduction in arid to semiarid environments iswater availability in the soil (Rodri´guez-Iturbe 2000;Flores & Briones 2001). Plant recruitment in these envi-ronments often occurs only in years with above-averagerainfall or in safe sites under the canopy of nurse-plants