Functional Ecology 2006 20, 58–66 58 © 2005 The Authors Journal compilation © 2006 British Ecological Society Blackwell Publishing, Ltd. Assessing the benefits of frugivory for seed germination: the importance of the deinhibition effect A. W. ROBERTSON,*† A. TRASS,* J. J. LADLEY‡ and D. KELLY‡ *Ecology, Institute of Natural Resources, Massey University, Private Bag 11222, Palmerston North, and ‡Biological Sciences, University of Canterbury, Private Bag 4800, Christchurch, New Zealand Summary 1. Many studies have examined the effects of frugivores on the germination of seeds of fleshy fruited plants. However, three key issues are rarely addressed: the need to measure germination of seeds in intact fruits; the effect of germination conditions on results; and the distinction between dead vs dormant seeds. 2. A literature review including 51 plant species from 28 families found that the often- measured scarification effect (germination of bird-defecated vs hand-cleaned seeds) is significantly smaller than the rarely-measured deinhibition effect (germination of hand- cleaned seeds vs those in intact fruits). 3. Both the literature review and new experimental data show that germination condi- tions affect germination. In particular, seeds in intact fruits have much lower germination percentages in Petri dishes than in the field. Poor germination from intact fruits in Petri dishes may be an artefact. 4. A field experiment with three New Zealand species showed variable effects of non- removal of the fruit pericarp. The retention of the pericarp had no effect on germina- tion in Nestegis cunninghamii; increased the proportion of seeds entering dormancy in Melicytus lanceolatus; and greatly increased seed mortality in Pennantia corymbosa. 5. Germination experiments must be designed carefully to evaluate accurately the risks for plants of frugivory mutualism failures. Key-words: fruit, germination experiment, germination inhibitors, gut passage, vertebrate seed dispersal Functional Ecology (2006) 20, 58–66 doi: 10.1111/j.1365-2435.2005.01057.x Introduction Vertebrate–fruit mutualisms can result in benefits to the plant, not only from the movement of seeds away from the parent (Schupp 1993; Willson & Traveset 2000), but also from changes to germination caused by the passage of a seed through the digestive tract of a vertebrate such as a bird (Krefting & Roe 1949; Rick & Bowman 1961; Ketring 1973; Van der Pijl 1982). These germination effects have been studied extensively. Frugivores can affect seed germination directly in three ways: (1) through scarification of the seed coat, which increases the per- meability of the coat to water and gases (scarification effect); (2) through removal of germination inhibitors by separation of the seeds from the pulp (deinhibition effect); and (3) through enhancement of germination and seedling growth from faecal material surrounding the seed (fertilization effect) (Traveset & Verdú 2002). However, two recent papers argue that much of the literature tests germination effects incompletely, or in ways that may introduce unintended effects (Kelly, Ladley & Robertson 2004; Samuels & Levey 2005). In this paper we present data to back up those arguments on three key points: first, the need to measure the ger- mination of seeds in intact fruits; second, the need to conduct such experiments in the field; and third, the need to discriminate between dead and dormant seeds at the end of trials. The first key point is the need to compare germina- tion of bird-dispersed seeds with that of seeds in intact fruits. Many studies have evaluated the germination benefit of dispersal by birds and other dispersal agents (Traveset 1998; Traveset & Verdú 2002). The majority of these studies test the scarification effect, by compar- ing the germinability of seeds that have passed through the gut of a vertebrate, with seeds that have been hand- cleaned (Samuels & Levey 2005). The difference in germination is generally interpreted as the benefit of gut passage, hence the dependence on vertebrates for germination enhancement. Traveset’s (1998) review, which included 183 plant species in 68 families, showed that scarification effects were small and inconsistent. Sixteen per cent of studies found that scarification had †Author to whom correspondence should be addressed. E-mail: a.w.robertson@massey.ac.nz