First Proof Of Batesian Mimicry in Two Plant Species!

Leaf shape, size, and colour are used by herbivores to identify sources of palatable foliage for food. It is possible, therefore, that an undefended plant might gain protection from herbivores by matching leaf characteristics of a chemically defended species. We demonstrate the use of a geometric morphometric approach to quantify spatial variation in leaf shape and size across populations of Pseudowintera colorata (Raoul) Dandy, and a putative Batesian mimic, Alseuosmia pusilla (Colenso) A. Cunningham. These are unrelated, sympatric species that, to the human eye, bear strikingly similar foliage. Using the Cartesian coordinates of leaf margins as descriptors of leaf shape, we found that in the chemically defended P. colorata, leaves were morphologically distinct from all of the neighbouring species except for the undefended A. pusilla. Alseuosmia pusilla individuals were more similar to neighbouring than to distant P. colorata, and 90% of leaf shape variation in the two species varied similarly across an altitudinal gradient. The data are consistent with Batesian mimicry, wherein the conspicuous characteristic of a defended model is replicated by an undefended mimic across its entire growing range. Our study provides the first detailed and powerful quantitative leaf shape evidence of leaf shape being matched between an undefended plant species to a chemically defended unrelated species across a shared growing range, and highlights the importance of using a spatially explicit morphometric method when investigating leaf shape, especially in relation to plant mimicry.

With the invention of lots of new items, we have found a lot of inventions that are transferred the world to a discover address now sample viagra newer look to the world. This bulk cialis continue reading that cause’s sexual frustration in a couple of years I could beat a lot of them at 50-point games, and 8-ball. But most importantly one should learn to take the initiative in the sexual life of the patient. cheap online levitra Impotence cialis online from canada can in some cases be a sign or symptom of some other serious problem.

Keywords: Batesian mimicry, herbivory, leaf shape, visual leaf signalling, phenotypic plasticity

Introduction

Many plant species exhibit morphological plasticity in response to ontogenetic, abiotic, competitive, and predatory pressures. These traits often fluctuate predictably across environmental gradients, suggesting that phenotypic plasticity confers a selective advantage to a plant in its natural habitat (Hovenden and Vander Schoor 2006; Royer et al. 2009). Leaf morphology, for example, assists in temperature regulation (Gurevitch and Schuepp 1990; Hegazy and El Amry 1998; Roth-Nebelsick 2001) and light interception (Horn 1971; Tsukaya 2005); it can also serve to minimise wind damage (Anten et al. 2010), to tolerate salinity (Sinclair and Hoffmann 2003), drought, or nutrient deficiencies (Cunningham et al. 1999; Leigh et al. 2011), or to deter predation by herbivores (Niemelä and Tuomi 1987; Brown and Lawton 1991). Within these broader trends, however, there is often considerable intraspecific variation in leaf shape even in the same environment (McDonald et al. 2003; Nicotra et al. 2011). The possible functional significance of such variation is usually far from clear.

Amidst this background of morphological heterogeneity, there are examples in which variation in the leaf shape, size, and colour of one species correlates closely with that in another, unrelated species. For example, the non-spiny-leafed Celmisia petriei and C. lyallii from New Zealand look like spiny members of the genus Aciphylla (Brown and Lawton 1991), whereas the nonstinging nettle like group of plants containing species from across several genera (Lamium album, L. purpureum, Lamiastrum galeobdolon, Ballota nigra, Galeopsis spp.), which are commonly known as dead nettles because of their lack of sting, resemble the defended European stinging nettle (Urtica dioica) (Brown and Lawton 1991; Wheeler 2004). In other cases, this correlation in resemblance only seems to occur in the presence of a particular species. In the heavily grazed desert–steppe transition zone of the northern Negev, Israel, the leaf form and overall habit of the hill ecotype of Iris atrofusca appears to be more similar to that of the unpalatable Asphodelus ramosus, than to the valley ecotype growing in the absence of A. ramosus (Shimshi 1979). Leaves on the woody vine Boquila trifoliolata of southern Chile, bear striking physical resemblance to those on their host tree; when portions of the vine traverse other hosts, their leaf shape changes accordingly (Gianoli and Carrasco-Urra 2014). Similarly, the leaves of some Australian mistletoes closely resemble those of their host plants (Barlow and Wiens 1977; Canyon and Hill 1997). In those instances the convergence in leaf shape has been hypothesised to present a potential fitness advantage by using mimicry to evade detection by herbivores. Leaf shape is known to influence the consumption, performance, and ovipositing choices of insect herbivores, and thus may provide a visual cue of a plant’s palatability (Rausher 1978; Mackay and Jones 1989; Rivero-Lynch et al. 1996; Campitelli et al. 2008).

Successful plant mimicry requires that the mimic and model converge in the perceptual world of an approaching herbivore such that it cannot easily distinguish between the two (Schaefer and Ruxton 2009). It needs to be demonstrated, therefore, that leaf shape, size, and colour of the putative mimic do indeed overlap with the ranges of those of the model, and that any morphological convergence is more evident between individuals where the two species co-occur. Previous studies of mimicry in vegetative shoots have relied on traditional measures such as leaf length, width, area, and numbers of lobes, which, although useful as broad descriptors of morphology, cannot resolve subtle differences in shape that might be used as a visual cue by an approaching herbivore. Here, we use a spatially explicit geometric morphometric approach, i.e., taking into account how leaf shape varies both within and between species over distance, to quantify leaf morphology in two species from a natural forest population in New Zealand. Geometric morphometrics provide a higher resolution of captured shape information than do the more traditional measurements (Rohlf and Marcus 1993; Klingenberg 2010).

Our focal plant, Alseuosmia pusilla (Colenso) A. Cunningham (Alseuosmiaceae), is an understory shrub found in conifer–broadleaf and beech forests of New Zealand. The species has attracted attention because leaf size, shape, and anthocyanin pigmentation make the leaves of individual plants at some locations indistinguishable, at least to the human eye, from those of an unrelated species, Pseudowintera colorata (Raoul) Dandy (Winteraceae), with which it is often sympatric (Fig. 1A) (Greenwood and Atkinson 1977; Dawson 1988; Atkinson and Greenwood 1989). The leaves of A. pusilla contain only a few antiherbivory compounds (Cambie and Parnell 1970) and are readily eaten by introduced ungulates (Greenwood and Atkinson 1977; Nugent et al. 2001); in contrast, those of P. colorata are rich in polygodial, a sesquiterpene dialdehyde that imparts a pungent taste and is a potent insect antifeedent (Asakawa et al. 1988). Leaves of P. colorata are also unpalatable to introduced herbivorous mammals (Forsyth et al. 2005), and, it has been suggested, may similarly have been avoided by moa (Greenwood and Atkinson 1977), the extinct flightless birds that are thought to have presented significant browsing pressure on New Zealand’s flora over the last 65 million years (Atkinson and Greenwood 1989; Wood et al. 2013). It has been postulated that at locations where the two plant species co-occur, A. pusilla may escape predation by effectively functioning as a Batesian mimic of P. colorata (Dawson 1988; Atkinson and Greenwood 1989). However, this hypothesis lacks any empirical evidence.

Fig. 1.Leaf morphology of Pseudowintera colorata and Alseuosmia pusilla. (A) Pseudowintera colorata (P.c) and A. pusilla (A.p) seedlings growing together under forest canopy (top), and in the open due to tree fall (bottom). (B) Selected examples to show the range of A. pusilla, and (C) P. colorata leaves across the altitudinal gradient. Scale bar = 20 mm. [Colour online.]

We hypothesise that: (i) the leaf shape of P. colorata is sufficiently distinct from those of neighbouring plant species such that its morphology might serve to signal its unpalatability; (ii) the leaf shape is shared exclusively by the undefended A. pusilla; (iii) variation in P. colorata leaf shape across an environmental gradient is matched by variation in foliar morphology of A. pusilla; and (iv) the similarity in leaf shape is greater the closer the two species are to one another.

Materials and methods

Plant materialLeaves from P. colorata and A. pusilla were collected from a 4.5 km stretch of track (40°53′30″S, 175°14′22″E, to 40°54′46″S, 175°15′37″E) through montane forest in the Otaki Forks region of the Tararua Forest Park, New Zealand. The collection area covered the full altitudinal range for the two species at this location (470–870 m), characterised by a shift from the lower conifer–broadleaf forest to the higher Lophozonia– and Fuscospora-dominated forest. Alseuosmia pusilla grew along the entire transect; P. colorata was most common above 720 m along the final 1.5 km of the transect, where it became the prominent understorey tree. Every P. colorata and A. pusilla within 2 m of one side of the track was sampled if the plants were between 15 cm and 2 m tall, and their locations were recorded using a Garmin 60 GPS (Garmin, Olathe, Kansas, USA; accuracy: ±2–10 m). The youngest fully expanded leaf was excised from a total of 211 P. colorata and 467 A. pusilla plants. When there were multiple branches, the youngest fully expanded leaf was removed from the highest branch.

In addition, we collected leaves from seven other tree and shrub species for which the leaf size and shape were similar to our focal species, and which were located within 1.5 m of P. colorata and A. pusilla plants. Between 60 and 130 individuals from each species were sampled; one leaf was collected from each individual from the closest branch facing the nearest P. colorata or A. pusilla plant. All leaves were refrigerated within 6 h of collection.

Leaf shape capturing

Digital images of the adaxial surface of the leaves were captured using a CanoScan 8400F flatbed scanner (Canon, Tokyo, Japan) at 600 dpi. Adobe Photoshop 5.0 (San José, California, USA) was used to re-create the original leaf shape by filling in any areas lost to herbivory. If the herbivory damage was too extensive for a confident reconstruction, the original leaf shape was estimated from comparison with that of remaining adjacent lamina. Leaf shape was quantified with LeafAnalyser 2.3.0 (Weight et al. 2008) using the Cartesian coordinates of 70 evenly distributed landmarks distributed around the margin of each leaf. The coordinates were exported into R 2.13.1 (The R Foundation for Statistical Computing) for statistical analysis.

Leaf colour

The reflectance spectra of green and red portions of a randomly chosen subsample of P. colorata (red, N = 5; green; N = 8), and A. pusilla (red, N = 4; green, N = 9) leaves were recorded using an USB2000 spectrometer (Ocean Optics, Dunedin, Florida, USA) and with an HL-2000-CAL light source. Diffuse reflectance was measured relative to that of a white Teflon standard. The fibre optics probe was fixed at 45° to the leaf’s adaxial surface, and spectra were measured at 0.2 nm intervals from 450–700 nm using the Ocean Optics SpectraSuite software.

Statistical analysisCartesian coordinates were transformed into partial Procrustes coordinates using the software package PAST 2.17c () to minimise differences in leaf orientation of the scanned images without scaling for leaf area. To quantify the similarity in overall leaf shape between P. colorata, A. pusilla, and the surrounding plant species at the field site, the partial Procrustes coordinate residuals (i.e., tangent space coordinates) sourced from PAST 2.17c were used in a discriminant function analysis (DA) with SPSS Statistics version 20 for Windows (IBM, Armonk, New York, USA). Using the full data set of all species as a training set, DA reclassified each individual leaf blindly into a species group based on how variation of leaf shape compared with the leaf shape of all individuals in the set.Because the sampling site spanned an altitudinal gradient of 400 m, we examined how variation in leaf shape between the two species altered with altitude. A principal component analysis (PCA) was run using the 70 partial Procrustes coordinates, reducing the number for each species into four principle components (PC) that explained the majority of shape variation for each leaf. Linear regressions for each PC as a function of altitude were compared for each species using ANCOVA.

To quantify spatial variation in leaf shape, we compared the Euclidean distances among A. pusilla and P. colorata plants on each PC plot with the geographical distances among them. To do this, the PC data were exported into R 2.13.1, whereupon the PC score of each A. pusilla or P. colorata was compared in turn with the PC scores of all individuals of the opposing species located within a 10 m radius. The radial distance from the focal plant was progressively increased in 10 m bands, and the process repeated until every individual of the opposing species across the sampling site had been compared. Using the log-likelihood function, we tested the likelihood of how similar each individual’s leaf shape was to those of the cohort of surrounding members of the second species at each 10 m interval as compared with those of the remaining members.

Results

Leaf morphologyThe leaf laminae of A. pusilla ranged from elliptic to cuneate, with an attenuate base and an acute apex, on average 77.3 ± 34.7 mm (95% CI) long and 24.0 ± 8.3 mm wide (Fig. 1B). They were submembranaceous, glabrous, and entire, borne on a petiole up to 10 mm long. Leaves of P. colorata were also typically elliptic but ranged to ovate; their laminae were coriaceous, glabrous, and entire or undulate, had a cuneate base and an acute to obtuse apex, and were generally larger than A. pusilla at 88.5 ± 36.6 mm long and 30.2 ± 12.1 mm wide, with a ≤10 mm petiole (Fig. 1C). Across the population, the two species overlapped in leaf dimensions; 62% of A. pusilla and 61% of P. colorata leaves were within 1 standard deviation (SD) of the other’s mean lamina length, and 51% of A. pusilla and 29% of P. colorata leaves were within 1 SD of the other’s mean leaf width. The two species shared similar reflectance spectra for the green as well as for the red portions of the leaf’s adaxial surface (Figs. 2A and 2B), indicating that there was negligible difference in lamina colour.

»View larger version

Fig. 2.Reflectance spectra of leaves. Proportions of reflected light from (A) green, and (B) red portions of Pseudowintera colorata (broken grey line) and Alseuosmia pusilla (solid black line) lamina. Data are the mean ± SE.

Shape analysisThe first four principle components of the PCA analysis accounted for 98.0% of the total leaf shape variation. Most (89.6%) of this was attributable to variation in leaf size and to size-associated shape differences represented by PC1 (Fig. 3A), with leaves of A. pusilla being generally smaller than those of P. colorata. In the remaining three PCs, P. colorata exhibited the wider variation in leaf shape, as evidenced by a greater spread in SD as compared with A. pusilla. PC2 accounted for 4.1% of the total variance in shape; it represented the morphological progression from an asymmetric elliptic for which lamina area was greater at the left- than at the right-hand side of the midrib as viewed from the leaf base on the adaxial surface (abbreviated hereinafter as left-biased), through to an asymmetric right-biased cuneate lamina (Fig. 3B). PC3 accounted for 3.4% of the variance and represented the opposite range to PC2, lamina shape varying from right-biased elliptic to left-biased cuneate (Fig. 3C). PC4 accounted for 1.0% of the variance, representing shapes from narrow linear to an elliptic (Fig. 3D).

»View larger version

Fig. 3.Principal components analysis (PCA) of leaf shape variation in Pseudowintera colorata and Alseuosmia pusilla. PCA reduced 98.1% of leaf shape variation between the two species into four principal components (PC1–4). For each PC, proportionate frequencies of P. colorata (grey bars) and A. pusilla (black bars) leaves are given for up to 3 standard deviations (SD) of the mean leaf shape (0). Leaf silhouettes show typical shapes, proportionately scaled, represented by each 1 SD on each PC. (A) PC1, accounted for 89.6% of total leaf shape variation; (B) PC2, 4.1%; (C) PC3, 3.4%; (D) PC4, 1%.

Shape discriminationThe leaf shapes of P. colorata and A. pusilla were morphometrically distinct from those of all seven other species growing in their vicinity; 97.3% of P. colorata and A. pusilla leaves (grouped together) were correctly identified by discriminant analysis. When the analysis was restricted to P. colorata and A. pusilla only, discriminant analysis incorrectly classified 7.1% of A. pusilla, of which most (5.4%) were classified as P. colorata. For P. colorata, 29.4% were misidentified as A. pusilla; none was misidentified as any other species (Table 1). Thus, 34.8% of A. pusilla and P. colorata were statistically indistinguishable from each other.

»View table

Table 1.Probability of correct identification of species by cross-validated discriminant analysis using the leaf margin landmarks.

Pseudowintera colorata exhibited a more even spread of individuals within ±3 SD from the mean leaf shape than did A. pusilla, especially in the characteristics represented by PC2 to PC4 (Figs. 3A–3D). Alseuosmia pusilla had a narrower range of shapes, characterised by a greater proportion of individuals approaching its mean leaf shape; thus, it was easier for the discriminant analysis to categorise the species correctly, explaining why a greater proportion of P. colorata was mistaken as A. pusilla rather than the other way round. Alseuosmia pusilla individuals that were found beyond ±3 SD from their mean leaf shape were closer to the mean shape of P. colorata than to their own mean shape (Table 1).

Altitudinal variation in leaf shapeLeaf size and shape represented by PC1 increased similarly in both species at higher altitudes (species identity: r² = 0.21, P < 0.001; slopes are parallel: df = 1, F = 0.28, P = 0.59; Fig. 4A). Leaves of P. colorata were constantly larger across the altitudinal gradient (intercepts differ: df = 1, F = 170, P < 0.001). However, some features of leaf shape varied with altitude differently between the species (all interactions between species and altitude in PC2, PC3, PC4 were statistically significant; P < 0.05). In A. pusilla, leaf shape shifted from left-biased elliptic to right-biased cuneate, whereas P. colorata shifted slightly in the opposite direction (PC2; Fig. 4B). For the attributes associated with PC3, the left-biased cuneate laminae of A. pusilla became less abundant at higher altitudes, replaced by right-biased elliptic leaves; there was no effect in P. colorata (Fig. 4C). Finally, the narrow linear laminae of A. pusilla (PC4) were replaced by more elliptic laminae, whereas there was no effect in P. colorata (Fig. 4D).

»View larger version

Fig. 4.Leaf shape of Alseuosmia pusilla and Pseudowintera colorata in relation to altitude. Leaf shape represented by (A) PC1, (B) PC2, (C) PC3, and (D) PC4, as illustrated in Fig. 3. Alseuosmia pusilla (crosses) and P. colorata (circles). Each point represents one leaf from an individual plant. The y-axes extend ±5 SD along each PC from the mean leaf shape, which is represented by 0. Linear regressions show altitudinal trends in leaf shape variation explained by each PC for A. pusilla (black line) and P. colorata (grey line). Leaf silhouettes are proportionately scaled and show typical shapes represented by ±3 SD and the mean leaf shape (0).

Shape comparison among neighboursThe probability that leaves of A. pusilla would be similar in shape to those of P. colorata was greater the closer the two species were to one another (Figs. 5A–5D). This probability fell rapidly at separation distances greater than about 200 m, irrespective of which criterion was used to describe shape. Similarity in leaf size and shape (represented by PC1) declined the most with increasing separation distance (Fig. 5A), followed by characters represented by PC2 (Fig. 5B) and then PC3 (Fig. 5C). Leaves that varied from narrow linear to elliptic (PC4 characters) were the least likely to change over distance (Fig. 5D). Patterns of similarity between P. colorata individuals and the surrounding A. pusilla did not change along the transect (see Figs. 6A–6D).

»View larger version

Fig. 5.Comparison of individual Alseuosmia pusilla principal component (PC) values to the neighbouring Pseudowintera colorata population. Log-likelihood scores for similarity in leaf shape represented by (A) PC1, (B) PC2, (C) PC3, and (D) PC4, comparing A. pusilla individuals to the surrounding P. colorata over distance. Decreases in log-likelihood scores indicate reduced similarities in leaf shape represented by the corresponding PC.

»View larger version

Fig. 6.Comparison of individual Pseudowintera colorata principal component (PC) values to the neighbouring Alseuosmia pusilla population. Log-likelihood scores for similarity in leaf shape represented by (A) PC1, (B) PC2, (C) PC3, and (D) PC4, comparing P. colorata individuals to the surrounding A. pusilla over distance. Decreases in log-likelihood scores indicate reduced similarities in leaf shape represented by the corresponding PC.

Discussion

There are five key findings from this study of spatial variation in leaf shape. (i) The chemically defended P. colorata is morphologically distinct from all surrounding heterospecifics except for A. pusilla (Table 1). (ii) The range of leaf shapes of P. colorata overlaps significantly with that of A. pusilla, meaning that 35% of the leaves of the two species cannot be statistically distinguished from one another (Table 1). (iii) Pseudowintera colorata and A. pusilla share similar lamina colours (Figs. 2A and 2B). (iv) Leaf shape in the two species varies similarly across an altitudinal gradient (Figs. 4A–4D). (v) Leaves of A. pusilla are more likely to resemble those of P. colorata the closer the two species are to one another (Figs. 5A–5D). These results add weight to the hypothesis that A. pusilla could be a Batesian mimic of P. colorata.

Batesian mimicry requires that distinguishable traits shared by model and mimic are identifiable by the predator and associated with unpalatability; the more conspicuous the signal, the more readily is this association learned (Cott 1940; Harvey 1983). For leaf shape to be effective as a visual deterrent, it would therefore need to be sufficiently distinct from those of surrounding species. Previous studies of putative Batesian mimics did not always address the question that the shared traits are exclusive to the putative model and mimic. Certain Australian mistletoes, for example, are remarkably similar in leaf colour and shape to those of their hosts (Barlow and Wiens 1977; Ehleringer et al. 1986), yet some of these similarities are also shared with other species in the surrounding plant community (Blick et al. 2012), and so it is likely that herbivores would use different features to discriminate between them (Canyon and Hill 1997). Like the mistletoes, the leaves of our putative model and mimic are very similar in shape and colouration; 35% of the leaves were statistically indistinguishable from one another. However, in contrast to the mistletoes, leaf shapes in our focal species are substantially different from those of all other sympatric trees and shrubs; discriminant analysis correctly classified 97% of A. pusilla and P. colorata when they were grouped together and compared with the surrounding species. Thus, there is a real possibility that leaf shape per se might be used as a visual deterrent in these species.

Leaf shape often changes along environmental gradients (McDonald et al. 2003). Thus, if mimicry is to be effective, both the model and mimic would need to respond similarly to any environmental factor that affects leaf shape. Along the altitudinal gradient spanned by our transect, changes in leaf size, and in leaf shape related to variation in leaf size, were similar for P. colorata and A. pusilla (Fig. 4A); these were PC1 features, accounting for 90% of the total shape variation shared between the two species (Fig. 3A). Moreover, the leaves of A. pusilla were morphologically most similar to those of neighbouring P. colorata within a radius of around 200 m, with a sharp decline in leaf similarity among plants that were further apart (Figs. 5A–5D). We do not know the degree to which genetic and environmental factors control leaf shape in these species, but it is evident that the two respond very similarly.

A foraging herbivore would most likely encounter potential food sources in succession rather than simultaneously, and would therefore rely on memory to associate leaf morphology with unpalatability. The accuracy of discriminating between successively viewed objects decreases rapidly, owing to inefficiencies in coding and retrieving of memories, as shown both in humans (Uchikawa and Ikeda 1981) and honeybees (Dyer and Neumeyer 2005). Thus, effective mimicry does not require that the mimics be exact replicas of their models; a plant would need only to resemble its protected neighbours with sufficient accuracy to account for the level of degraded memory. Rather than a general similarity in leaf morphology, the similarity shared between A. pusilla and its immediate P. colorata neighbours, which is maintained across the shared growing range, would further reduce a foraging herbivore’s ability to discriminate between them.

We do not know which of the potential herbivores may have been important in driving the evolution of leaf morphology in A. pusilla. The terrestrial fauna was historically dominated by moa, which were generalist herbivores consuming a wide variety of plants (Wood et al. 2008, 2012), but there is no evidence to suggest that they were the main selective pressure in this system. Little, too, is known about the insect herbivores of A. pusilla. Forest populations of A. pusilla often show evidence of damage by Lepidopteran larvae, but it is not known whether these and (or) other herbivorous invertebrates attack or are deterred by P. colorata. Insects have poor spatial acuity compared with larger vertebrates (Giurfa et al. 1996), but once close enough to a plant they may similarly resolve small objects to recognise patterns and shapes (Giurfa and Menzel 1997). Thus, leaves might well have been used by not only moa, but also insect herbivores for identification purposes.

Unequivocal proof of mimicry requires evidence that: (i) both the mimic and a known model are attacked by the same herbivore; (ii) the herbivore is unable to distinguish between them; and (iii) this confusion increases the fitness of the mimic. To date, no study has satisfied these three criteria in plants (Schaefer and Ruxton 2009, 2011). However, our study provides strong evidence that leaf shape is precisely matched between an undefended plant species and a chemically defended unrelated species across a shared growing range, and therefore represents an important first step at establishing Batesian mimicry between A. pusilla and the putative model, P. colorata. We highlight how a spatially explicit morphometric analysis provides a powerful tool to study visual Batesian mimicry in plants.