To obtain a nutritionally adequate diet, generalist herbivores are predicted to forage on a variety of plant species (Westoby Reference WESTOBY1974). Most free-ranging herbivores, however, forage non-randomly in the presence of a broad assortment of plants (Marsh et al. Reference MARSH, WALLIS, ANDREW and FOLEY2006). Thus, while nutrient concentrations in leaves may drive food selection in some herbivorous species (Pérez-Harguindeguy et al. Reference PÉREZ-HARGUINDEGUY, DÍAZ, VENDRAMINI, CORNELISSEN, GURVICH and CABIDO2003), other factors such as toxic secondary compounds (Freeland & Janzen Reference FREELAND and JANZEN1974), foliar water content (Poorter et al. Reference POORTER, VAN DE PLASSCHE, WILLEMS and BOOT2004) or leaf mechanical strength (Kitajima & Poorter Reference KITAJIMA and POORTER2010) can also influence selection. In herbivorous lizards, little is known about what prompts their foraging decisions, although it has been shown that they can exhibit specific food preferences under natural conditions (Dearing & Schall Reference DEARING and SCHALL1992, van Marken Lichtenbelt Reference VAN MARKEN LICHTENBELT1993). Understanding food-selection behaviour in herbivorous lizards, a specialized group comprising approximately 1% of lizard species (Cooper & Vitt Reference COOPER and VITT2002), is important given their role as structural modifiers of vegetation communities (Auffenburg Reference AUFFENBERG, Burghardt and Rand1982) and as potentially significant seed dispersers (Moura et al. Reference MOURA, CAVALCANTI, LEITE-FILHO, MESQUITA and MCCONKEY2015).

In this study we examined the influence of leaf traits on food selection of a free-ranging herbivorous lizard, the Exuma rock iguana (Cyclura cychlura figginsi Barbour 1923), by quantifying selection using faecal examinations, and relating it to leaf traits (leaf area and density), chemical concentrations of nutrients and fibre fractions. Since it is expected that generalist herbivores require a variety of elements to fulfil their nutritional demands, we hypothesized that the iguana would not select plants based on any singular foliar nutrient concentration but would prefer leaves with overall higher nutrient concentration. Additionally, high material strength, such as those conferred by fibrous carbon polymers, can decrease food digestibility and minimize foraging efficiency (Clissold et al. Reference CLISSOLD, SANSON, READ and SIMPSON2009). For example, hemicellulose provides pliable cross-links between cellulose microfibrils, whereas cellulose influences the directional tensile strength of primary and secondary cell walls, and lignin provides permanent and rigid cross-linkages within the secondary cell wall (Kitajima et al. Reference KITAJIMA, LLORENS, STEFANESCU, VARGAS TIMCHENKO, LUCAS and WRIGHT2012). Thus, we expected iguanas to select leaves with low tissue density and low per cent fibre fractions.

The Exuma rock iguana is a large-bodied (max. snout–vent length and body mass: 58.0 cm and 7.94 kg, respectively) herbivorous lizard inhabiting cays in the Exuma Islands chain, Bahamas. We conducted research on Gaulin Cay (24°07′18.32′′N; 76°24′19.57′′W), a 13.6-ha island with a vegetation community consisting of 30 species. In May 2010, we captured iguanas using fish-landing nets and by noose. All captured iguanas were measured (snout–vent length, body mass) and sexed by cloacal probing for hemipenes. Faecal samples were collected from 30 iguanas primarily during processing. If iguanas did not defecate during processing, they were kept overnight in opaque cloth bags and samples collected the next morning. Faecal samples were examined and visible plant remains identified to species using Correll & Correll (Reference CORRELL and CORRELL1982).

We estimated plant species abundance using a 1-m² quadrat placed every 10 m along nine 100-m-long transects. We estimated preferences for each food type by calculating electivity values defined by Ivlev (Reference IVLEV1961): Electivity = [r(i) − p(i)]/[r(i) + p(i)], where r(i) is the proportion of food type i in the diet and p(i) is the proportion of food type i in the environment. The result is a metric ranging from −1 (food considered strongly avoided) to 0 (food taken in its proportion in the environment) to + 1 (food considered strongly preferred).

We focused our dietary analysis on leaves because seasonal fruits are reported eaten regardless of plant species whereas leaves are not (Auffenburg Reference AUFFENBERG, Burghardt and Rand1982). Leaves also represent a consistent, available food supply during our restricted sample period and remain intact through the digestive system. To study physical and chemical properties of foliage, we selected the 10 most common plant species and collected leaf samples from five plants per species. We calculated mean leaf area from 10 scanned leaves per species using Scion Image software (Scion Corporation, USA). We measured leaf thickness from the same fresh samples using a micrometer. Leaf volume was calculated as mean leaf area × mean leaf thickness. Samples were then dried at 60°C and mean leaf density calculated as leaf dry weight/leaf volume. We used leaf density as a proxy for leaf material toughness (resistance to fracture per unit fracture area) because they are positively correlated (r = 0.54) in 197 species of tree and shrub species (Westbrook et al. Reference WESTBROOK, KITAJIMA, BURLEIGH, KRESS, ERICKSON and WRIGHT2011) and because leaf density is a good indicator of physical defence (Kitajima & Poorter Reference KITAJIMA and POORTER2010).

We measured nutrient and fibre concentrations using additional foliage from the collected samples. We measured total per cent N, with an elemental analyser (Costech Analytical, USA), and per cent P using ash digestion (Jones & Case Reference JONES, CASE and Sparks1996) followed by colorimetric determination of ortho-phosphate using a spectrophotometer microplate reader (Bio-Tek Instruments Inc., USA). We measured per cent Ca, K, Mg using ICP spectroscopy. As an indicator of digestibility, we measured fibre fractions on leaves (per cent hemicellulose, cellulose and lignin), using a series of increasingly aggressive extractants in a fibre analyser system (ANKOM Technology, USA; Alvarez-Clare & Kitajima Reference ALVAREZ-CLARE and KITAJIMA2007).

To test for differences in physical and chemical properties among plant species, we conducted Kruskal–Wallis non-parametric tests. We evaluated the relationship among species means using pairwise Spearman correlations. We conducted linear regressions to test if mean foliar physical and chemical traits were good predictors of selection. We included biologically important ratios (Mans & Braun Reference MANS and BRAUN2014) in our analyses but do not report results as none were statistically significant.

We identified one to three plant species in each faecal sample (mean ± SD = 1.6 ± 0.7 species) and did not detect a significant relationship between iguana size and number of items in faecal samples (r² = 0.04, P = 0.40). All measured morphological and chemical leaf traits varied significantly among plant species but only N and P concentrations (Spearman ρ = 0.78, P = 0.01), and N and K concentrations (Spearman ρ = 0.82, P < 0.01) were correlated across species (Table 1). In addition, from all measured leaf traits only leaf density was a predictor of selection as iguanas preferred leaves with lower tissue density (r² = 0.38, P = 0.05).

Table 1.

Mean (± SE) measured leaf traits for Bahamian plants listed by increasing preference (e.g. electivity index (EI)) by the Exuma rock iguana. Included are non-parametric Kruskal–Wallis H statistics comparing leaf traits across species. Plant species follow nomenclature of Correll & Correll (Reference CORRELL and CORRELL1982). Hemi, hemicellulose; Cellul, cellulose; * 0.05 > P > 0.001; ** P < 0.001.

^a SE is zero because only one plant was used to obtain this measurement.

Our results demonstrate substantial variability in nutritional traits among potential food types for iguanas, but no apparent selection toward any specific nutritional trait in leaves. As predicted, our data suggest that food selection over the short term is not driven ultimately by a singular nutritional component. Similar findings have been reported for other herbivorous lizards regarding nutritional (Blázquez & Rodriguez-Estrella Reference BLÁZQUEZ and RODRÍGUEZ-ESTRELLA2007) and caloric (Auffenberg Reference AUFFENBERG, Burghardt and Rand1982) values. Together, these results support the nutritional constraint model, which purports that no one plant species can satisfy nutritional demands of a herbivore generalist (Westoby Reference WESTOBY1978) because plant tissues are relatively low in essential nutrients and digestible energy compared with those of animals (Robbins Reference ROBBINS1993).

Our results provide evidence that leaf material toughness (i.e. leaf density) can influence food selection in herbivorous iguanas over the short term and are consistent with studies conducted on invertebrates (Coley Reference COLEY1988) and primates (Teaford et al. Reference TEAFORD, LUCAS, UNGAR and GLANDER2006). Although we did not find a direct correlation between selection and the individual fibre fractions (% hemicellulose, % cellulose and % lignin), this is not surprising given that leaf toughness is not conferred exclusively by the chemical composition of the leaf but by the structural combination of these heterogeneous materials (Kitajima et al. Reference KITAJIMA, LLORENS, STEFANESCU, VARGAS TIMCHENKO, LUCAS and WRIGHT2012). Leaf density has been shown to be a good predictor of overall leaf toughness across multiple species (Kitajima & Porter Reference KITAJIMA and POORTER2010, Westbrook et al. Reference WESTBROOK, KITAJIMA, BURLEIGH, KRESS, ERICKSON and WRIGHT2011), and our study supports this.

Toughness in leaves can affect iguanas by slowing foraging efficiency and by decreasing food digestibility. For any two given foods (other features of their loading in the mouth being similar), a tougher food will require more work per area to fracture during chewing (Wright & Vincent Reference WRIGHT and VINCENT1996). Investigations focusing on prey processing behaviour and jaw mechanics suggest that reptiles chew (Ross et al. Reference ROSS, BADEN, GEORGI, HERREL, METZGER, REED, SCHAERLAEKEN and WOLFF2010; but see Fritz et al. Reference FRITZ, HUMMEL, KIENZLE, STREICH and CLAUSS2010) and iguanas often bite repeatedly once a piece of leaf has been sheared, or when leaves have been pulled away at the petiole. Digestion by microbes is facilitated by cuts or fractures in the cuticle (Bjorndal et al. Reference BJORNDAL, BOLTEN and MOORE1990), and thus it is conceivable that digestion efficiency is increased via punctures from repeated bites that are deeper and more severe in a leaf that is relatively less tough. Additional research is needed to support this supposition (Bjorndal et al. Reference BJORNDAL, BOLTEN and MOORE1990).

Our results offer a basis for future testing in herbivorous lizards and should be expanded to include influences of varying leaf morphological, chemical, and biomechanical traits resulting from seasonal effects and phenological stages of leaves. Iguanas consume seasonal fruits; therefore, expanding this study to consider fruits and flowers would further advance our understanding of food selection criteria in herbivorous lizards. Nonetheless, the results from this study are the first to demonstrate that physical leaf traits can influence food selection in a true herbivorous lizard and offer a testable hypothesis for further study.