Anuran–prey selection might be mediated by traits, either by mismatches in predator and prey traits (preventing interactions) or by predator selection of prey traits (encouraging interactions). These effect traits could be summarized in two contrasting foraging strategies: “active” and “sit‐and‐wait” foragers. We evaluated whether anurans could be classified into groups of species sharing traits associated with their diet, and what is the relation between particular effect traits of anurans and their prey. We collected anurans and identified their stomach contents once during dry, minor, and major rain seasons in six dry forest sites in the Colombian Caribbean. For each of the 19 anuran species and 436 prey items, we registered six effect traits. We applied RLQ and fourth‐corner methodologies to relate predator and prey traits through their interaction matrix. Predators were assigned to five groups according to their differences in locomotion, body shape, proportion of the jaw width, mode of tongue protrusion, and strata preferred. Regarding preys, species were assigned to four groups according to their gregariousness, body shape and hardness, defensive traits, and mobility. Body size of both, predators and prey, had a minor contribution in the group assignment. We found that predators using active search target low‐mobility preys, whereas species using sit‐and‐wait strategy target highly nutritive prey that are difficult to manipulate. By linking amphibian diet with foraging strategies, we hope to contribute to the understanding of mechanisms behind anuran–prey food web patterns and to build more realistic models of functional response to changing environments.

We collected anurans and identified their stomach contents once during dry, minor and major rain seasons in six dry forest sites in the Colombian Caribbean. For each of the 19 anuran species and 436 prey items, we registered six effect traits. We applied RLQ and Fourth-corner methodologies to relate predator and prey traits through their interaction matrix.

We sampled six sites within the tropical dry forests of the Colombian Caribbean. The sites were located between 11°5'26.68" N - 72°38'38.84" W and  8°47'35.24" N - 76°19'43.31" W, and separated, on average, by a linear distance of 115.3 km. Local elevation was less than 500m above sea level, and the region included both forest remnants and intervened areas. We surveyed each site during the dry season (February-March), at the onset of "minor rains" (May-June), and during the rainier season (October-November). Each visit consisted of two-hour searches for five nights between 1900 h and 2300 h. During each search, we encountered anurans through visual surveys, collected every recorded specimen, and preserved the specimens according to Angulo et al. (2006). We collected a minimum of five and a maximum of 146 (mean ± SD = 58.2 ± 39.1) individuals per species, focusing on adult and sub-adult specimens only. With that, we avoided the interference from changes in the diet that take place when juveniles become sub-adults. It must be noted that the same analyses excluding sub-adult specimens gave similar results (Blanco-Torres pers. comm.).

In the laboratory, we screened the stomach content of each specimen collected, registering the number of individuals per prey item. Arthropod preys were identified at the level of species or morphospecies by specialists at the Universidad Nacional de Colombia and Universidad del Atlántico-Colombia. Voucher specimens were deposited in Universidad del Atlántico’s museum

We selected and quantified six predator effect traits that are known to influence food intake by species, as per Trochet et al. (2014), Vitt & Caldwell (2014) and Cortés Gómez et al. (2015) (Table S1). Morphological traits included: 1) body size (S), as the snout-vent length (range: 7-131 mm), 2) body / jaw size ratio (SJ, range: 1-6.03), calculated as the proportion of the jaw width (as the horizontal length of the open mouth cavity measured from collected specimens) relative to the body length, 3) body shape (BS), assigned to the more similar geometric shape of the contour of the predator body, and classified according to four categories: BS1: pointed-elongated; BS2: ovoid; BS3: elongated; BS4: rounded-pointed (Fig. S1), 4) locomotion mode (LM), which classified species as burrowers (LM1), walkers (LM2), hoppers (LM3), terrestrial jumpers (LM4) or arboreal jumpers (LM5) after Reilly & Jorgensen (2011), 5) tongue protrusion mode (T), classified as mechanical expulsion of the tongue through muscle contraction (K1), inertial elongation (K2), or hydrostatic elongation (tongue lengthened through volume limitations imposed by the constant volume of an hydrostatic muscle, K3), after Nishikawa (1999), 6) preferred strata classified as terrestrial (H1), arboreal (H2), or cave dwelling (H3), based on our own surveys and confirmed through bibliography (e.g., Renjifo & Lundberg 1999, Cuentas et al. 2002, Blanco-Torres et al. 2019).

We also quantified six prey effect traits: 1) body shape (bs), assigned to the more similar geometric shape of the contour of the prey body, according to the following categories: rectangular (bs1), circular (bs2) or ovoid (bs3; Fig. S2), 2) body hardness (bh), classified according to the level of sclerotization as soft (bh1), medium (bh2) or chitinous (bh3), 3) size (s), measured from the specimens obtained from the stomach content analysis (range:1-1600 mm³), 4) mobility (m), based on the presence and type of legs, and the presence of wings, the prey specimens were classified as larvae (m1), walkers (m2), jumpers (m3) and flyers (m4), after Speight et al. (2008) and Triplehorn & Johnson (2005), 5) defense type (d), which was classified as: chemical (d1), mechanical (d2), or inexistent (d3), based on the most frequent defense type within the genus or, in most cases, family level, after Speight et al. (2008) and Triplehorn & Johnson (2005), and 6) gregariousness (g), according to the social structure of species, genus or family, classified as gregarious (g1) or not gregarious (g2), after Speight et al. (2008) and Triplehorn & Johnson (2005). Whenever the prey specimens were broken or partially digested, they were compared to intact specimens of similar morphotype to confirm the trait value assigned.