Frugivore behaviour determines plant distribution: a spatially-explicit analysis of a plant-disperser interaction

• Published in: Ecography (Copenhagen) 2012-02, Vol.35 (2), p.113-123
• Main Authors: Rodríguez-Pérez, Javier, Wiegand, Thorsten, Santamaria, Luis
• Format: Article
• Language: English
• Subjects: Activity patterns; Animal and plant ecology; Animal behavior; Animal scales; Animal, plant and microbial ecology; Animals; Biological and medical sciences; Daphne; Fundamental and applied biological sciences. Psychology; General aspects; Habitat preferences; Lizards; Mature plants; Predictive modeling; Seed dispersal; Shrubs; Spatial distribution; Wildlife habitats
• ISSN: 0906-7590; 1600-0587
• DOI: 10.1111/j.1600-0587.2011.06930.x

The spatial distribution of plants (and other primarily sessile organisms) depends on the interplay between their ecological requirements and the spatial template set before, during, and after the dispersal process. In the case of animal-dispersed plants, the spatial characteristics of animal behaviour during the seed dispersal process are likely to leave a lasting imprint on plant distribution. Here, we hypothesize that the activity patterns of the frugivorous lizard Podareis lilfordi directly influence the spatial distribution of the fleshy-fruited shrub Daphne rodriguezii. To evaluate this hypothesis, we first analysed lizard activity, following radio-tracked lizards during the plant's fruiting period, and identified its main determinants at several spatial scales of habitat aggregation (from 12.5 to 150 m). We hypothesised that lizard activity depends on differential habitat features explaining its territory use plus habitat preferences associated with each movement bout. In a second step, the most important determinants of lizard activity plus the variables describing habitat structure were used to predict the presence of adult and juvenile plants. Predictability of lizard activity (based on AUC and Pearson regression coefficients) was higher at broad spatial-scales of habitat aggregation (75 m). The two best predictors of lizard activity were the habitat features of and the distance to the core area (defined as the area enclosing the 0.50 cumulative probability of lizard locations). Plant presence was best predicted by models based on a combination of lizard activity and habitat features at local spatial scales (1.5 m). Best models included habitat features and lizard activity for adult plants, and local-scale habitat features, the proximity of adult plants and lizard activity for juveniles. In both cases, most plants (50-60%) were located at Optimal sites' (both favourable for lizards and with adequate habitat features), whereas a small fraction of them (3-10%) were located at dispersal-limited sites (i. e. with adequate habitat features but suboptimal for lizards). Our results thus suggest that the interplay between lizard activity and local habitat features determines the spatial patterns of juvenile-plant presence and leaves a lasting signature on adult-plant distribution.