Functional Ecology 2015 doi: 10.1111/1365-2435.12479 The relationship between microhabitat use, allometry and functional variation in the eyes of Hawaiian Megalagrion damselflies
Jeffrey A. Scales*,1,2 and Marguerite A. Butler2
1Department of Integrative Biology, Univeristy of South Florida, Tampa, FL 33620, USA; and 2Department of Biology, University of Hawaii at Manoa, Honolulu, HI 96822, USA
Summary 1. The evolution of visual systems is guided by visual requirements imposed by the environ- ment, the size of the animal’s eyes, and physical limitations imposed by the resolution-sensitiv- ity trade-off. Given a particular eye surface area, resolution and sensitivity cannot be simultaneously maximized: gains in resolution, the ability of the eye to detect detail, will come at the cost of sensitivity, the ability to capture photons, and vice versa, without an increase to eye size. How this constraint interacts with ecology and whether it allows the fine-tuning of the visual system to smaller scale habitat heterogeneity remains an understudied question in visual ecology. 2. Here, we use closely-related species of damselflies in the Hawaiian genus Megalagrion which differ in ecology to test whether variation in the resolution-sensitivity trade-off is the evolution- ary result of scaling or differences in microhabitat use. We use regression analyses and phylo- genetic comparative methods to examine the effects of size and microhabitat use on traits related to light sensitivity and visual resolution. 3. We find that eye size is tightly associated with body size in these damselflies, but other visual morphology traits related to light sensitivity and resolution are associated with micro- habitat type. Furthermore, size and morphology relationships vary across microhabitats, and performance related to resolution tends to be more conserved than to variation in light sensitivity. 4. Additionally, smaller species in visually challenging microhabitats have more regionalized eyes than species with larger eyes in open, well-lit areas. Thus, regionalization of the eye allows a decoupling of size and morphology/performance so that even small insect species can exploit visually challenging habitats. 5. These results suggest that variation in visual performance results from changes in eye geom- etry as well as size. These morphological changes are likely adaptive to differences in micro- habitat, indicating that variation in microhabitat use, even at small scales, can play an important role in the evolution of visual systems.
Key-words: adaptation, allometry, damselfly, compound eyes, microhabitat, ommatidia, visual ecology, vision, visual morphology
sensitivity is the ability to capture light. Resolution and Introduction sensitivity cannot be simultaneously maximized due to The eyes, ‘windows to the world’, have long-inspired evo- their opposing dependence on surface area and, as such, lutionary functional study (Walls 1942), uniting a wonder- represent key constraints that guide the evolution of visual fully complex functional system with strong selective systems (Stavenga 1979; Land 1989; Land & Nilsson pressures imposed by environmental variation. Spatial res- 2002). Furthermore, smaller animals are expected to be olution is the ability to distinguish objects, whereas light closer to the functional limit for both parameters. Appo- site to this trade-off is the notion that eye size is con- strained by physiological costs (Laughlin et al. 1998). *Correspondence author. E-mail: [email protected]
© 2015 The Authors. Functional Ecology © 2015 British Ecological Society 2 J.A. Scales & M.A. Butler
Indeed, eye size follows strong allometric scaling in many (D/), two parameters which contribute substantially to taxa (Brooke, Hanley & Laughlin 1999; Kiltie 2000; Hall visual performance (Sherk 1978). The genus Megalagrion 2007; Hall & Ross 2007). Therefore, variation in the resolu- is also exceptional in habitat breadth, covering the full tion-sensitivity trade-off and its relationship to size are range of possible damselfly breeding habitats from streams expected to evolve in response to differences in lifestyle or to waterfall seeps to phytotelmata (pockets of water held habitat. These design principles have received some com- by plants) to terrestrial leaf litter (Polhemus & Asquith parative study, (but often not in a phylogenetic context, 1996). Importantly, these vary in light levels (from full sun Jander & Jander 2002; Kapustjanskij et al. 2007; to very dark) and structural complexity (highly cluttered Rutowski, Gislen & Warrant 2009), or only with respect to to open). All Megalagrion have territorial mating systems the most extreme light habitat transitions (e.g. diurnal vs. and apparently lack female mate choice (Polhemus & nocturnal or crepuscular shifts; Thomas et al. 2006; Hall Asquith 1996), presenting a more direct evolutionary con- 2007; Hall & Ross 2007; Schmitz & Wainwright 2011). nection between sensory system characteristics, male trait Therefore, it is not known whether allometric constraints properties and signalling behaviour. or adaptation dominates in the evolutionary diversification How resolution and sensitivity are potentially tuned to of visual systems in complex habitats. ecological variation is particularly convenient to study in Visual demands can vary widely across environments the compound eyes of insects, which have external retinas (Endler , 1990, 1993; Thery 2001; Warrant 2004). Accord- allowing easy characterization of visual geometry. Apposi- ingly, visual system adaptations are well known from tion compound eyes have a modular design comprised of animals which inhabit extreme visual environments such as ommatidia or individual sensory units that contribute sin- the dim, blue-shifted light of the deep sea (reviewed in: gle ‘pixels’ to the viewed image (Fig. 1). Increasing light Warrant & Locket 2004) or animals active during the low- sensitivity therefore requires an increase in the surface area light periods at dusk or night (Endler 1991; McIntyre & of the lens of each ommatidium, or facet diameter (D). If Caveney 1998; Warrant 2004; Rosenthal 2007). Land- eye size does not correspondingly change, the increases in scape-level differences in terrestrial habitats such as forests (D) result in fewer ommatidia per eye surface area and vs. grasslands that differ in light regimes can likewise pro- therefore reduced resolution capabilities (which is mea- duce visual adaptations (Bauer et al. 1998; Veilleux & sured by interommatidial angle D/, the spatial sampling of Lewis 2011). However, the smaller scale differences offered the eye). Many insects alleviate this trade-off by possessing by heterogeneities within a single habitat or ‘microhabi- a specialized region called an acute zone (analogous in tats’ have received far less study (Marchetti 1993; Endler function to the fovea of vertebrates), where interommati- & Thery 1996; Marshall 2000; Endler & Mielke 2005). Sig- dial angle is decreased, while the lens facet diameter is naling animals are known to choose optimal locations simultaneously increased, resulting in better visual perfor- within a habitat to improve the transmission of their visual mance at the expense of performance elsewhere in the eye signals (Mallet & Gilbert 1995; Thery & Vehrencamp (Stavenga 1979; Land 1989). Specifically, we test whether 1995; Long & Rosenqvist 1998; Gomez & Thery 2004), variation in eye size, D and D/ is explained by allometry, but few examples are known of signal receivers doing the by microhabitat effects, or their interaction (for alternative same, or of visual systems tuned to these smaller differ- hypotheses see Fig. 2). Here, we define allometry as a scal- ences in light quality (Endler 1992; McDonald 1995; Leal ing relationship with body size, with isometry as a special & Fleishman 2002). Yet, specialization to microhabitats case. Isometry is observed when there is a one-to-one may provide ample ecological opportunity for the adaptive increase in length between the different body parts being evolution of animal eyes. compared. Here, we examine how overall body size and variation in microhabitat influence eye morphology and visual per- Materials and methods formance in the endemic Hawaiian damselfly genus Megalagrion. Megalagrion damselflies are ideal for study- STUDY SPECIES ing visual system design, as it contains 16 abundant and extant species, which have entered new breeding habitats Adult male individuals of 13 species of Megalagrion damselflies (and associated light environments) at least five times. In were collected from the wild using hand nets from the islands of Molokai, Maui, Oahu and Kauai (Table 1) from 2010 to October addition, they show nearly twofold variation in body size of 2012. All individuals within a species were collected from the and have evolved bright body coloration numerous times same island. We focused on adult males in this study for two rea- (Polhemus & Asquith 1996; Polhemus 1997). Damselflies sons. First, females are difficult to collect (often located only when are aerial predators and have large eyes which appear to in tandem with a male). Secondly, male odonates use visual signal- have reached a point of physical limitation, occupying the ling in establishing territories and finding mates (Corodoba- Aguilar 2008) which may result in a tighter link between visual majority of the head’s surface (Sherk 1978). Odonates characteristics and microhabitat. Only for M. koelense were adult (damselflies and dragonflies) have the highest visual resolu- females included in eye morphology data (the two males and two tion of any insect group, but also show unusually high females did not show any differences). All morphological measure- variation in facet diameter (D) and interommatidial angle ments were conducted within 3 days of capture.
© 2015 The Authors. Functional Ecology © 2015 British Ecological Society, Functional Ecology Microhabitat use and damselfly vision 3
Facet
Crystalline Increased
diameter cone (FD) } light sensitivity Interommatidial angle Δφ
Rhabdom (b)
Increased Fig. 1. Variation in the design of com- resolution } Ommatidia pound eyes. (a) Light enters the ommatidia; individual sensory units that make up a (a) Smaller compound eye, through a facet lens and is focused on a light receptor rhabdom. (b) Apposition Large facet diameters increase light sensi- compound eye Δφ tivity by creating a larger aperture for incoming light, but this will reduce resolu- tion by increasing interommatidial angle (c) (D/). (c) A fine-grained image requires numerous ommatidia to sample the viewed Regionalization scene. Small D/ indicate that the omma- tidia are arranged nearly in parallel. (d) Regional variation of the eye can produce an area of high light sensitivity and resolu- tion. Large ommatidia and small D/ in the Smallest high-resolution zone are achieved by a Δφ localized flattening of the eye, but requires greater curvature (and lower resolution) elsewhere. (e) Eye photograph of Megalag- rion nigrohamatum nigrolineatum, showing the hexagonal array of lenses and large var- (e) (d) iation in D.
appear to be similar across islands however, terrestrial species are MICROHABITAT TYPES found only on Oahu. The five microhabitat types used in this study (stream, pool, seep, plant and terrestrial) are those recognized by field workers and MORPHOLOGICAL MEASUREMENTS taxonomists (Polhemus & Asquith 1996, Table 1). Males of this genus are generally territorial, guarding preferred oviposition sites Gross morphology measurements were comprised of two measure- at specific structural microhabitats where visual performance may ments: Thorax length (THORAXL), the proxy for body size, was be important. We use ‘microhabitat’ here to indicate small loca- the distance from the junction of the second leg and the thorax to tions within a larger habitat that are preferred by a particular spe- the base of the hindwing. Eye height (EYEHT), the proxy for eye cies. Microhabitats may define the ‘niche’ of species within a size, was the diameter of the eye along its vertical axis and community, not only in terms of where it is found, but also corre- measured at the tallest part of the eye. All measurements were per- lating with strategies for feeding, locomotion or other aspects of formed with ImageJ (Rasband 2012) on lateral photographs of functional biology (Elton 1927; Gause 1934; Hutchinson 1957; individual damselflies including a size standard (taken with a MacArthur 1958; Williams 1983). Canon EOS 5D digital camera, Canon, Melville, NY, USA). Each ‘Stream’ species are those that commonly use open areas in morphological variable was checked for deviations from normality large, swift streams. ‘Pool’ species frequently occur along stream using qq-plots in the R computing environment (R Core Team margins, pools and ponds. Pool species use water that is slower 2013). We log-transformed body size and eye size to facilitate moving and often associated with vegetation. ‘Seep’ species are interpretation of scaling relationships. As the data were statisti- associated with the wet vertical surfaces of stream valleys and bed- cally well-behaved (met the assumptions for parametric statistical rock walls in deep forests, as well as bogs. The naiads of seep spe- tests), the remaining variables were analysed on their natural cies are not found in streams, but under wet moss covering the scales (i.e. untransformed) for ease of interpretation. walls with permanent water seepage. ‘Plant’ species breed in water pockets in the leaf axils of plants with a bromeliad growth form and are found in forests or bogs, typically away from streams. INTEROMMATIDIAL ANGLE AND FACET DIAMETER Finally, the single ‘terrestrial’ species breeds in the leaf litter of the MEASUREMENTS native Uluhe fern and can be found in the forest, along trails and ridge tops (Williams 1939). Megalagrion oahuense is one of the We mapped variation in facet diameter and interommatidial angle very few damselfly species in the world known to be fully terres- across the eye as correlates of visual performance. Whereas spa- trial. These five microhabitats are present on all islands and tial resolution is determined by interommatidial angle, several
© 2015 The Authors. Functional Ecology © 2015 British Ecological Society, Functional Ecology 4 J.A. Scales & M.A. Butler
parameters contribute to light sensitivity (Land & Nilsson 2002) utor to variation in light sensitivity will be differences in facet which can vary between species. Along with facet diameter, differ- diameter (with the caveat that other traits contribute to light sensi- ences in focal length, rhabdom diameter and the ratio between the tivity and to the extent that they vary may also influence visual two may all play a significant role in determining light sensitivity. abilities), an assumption made by previous comparative studies of However, because the species included in our study are very clo- compound eyes (Jander & Jander 2002; Rutowski, Gislen & sely related (separated by between 0 5–10 Myr Jordan, Simon & Warrant 2009). Polhemus 2003), it is reasonable to assume that a primary contrib- We measured facet diameters (D) and interommatidial angles (D/) using the pseudopupil technique (Horridge 1978; Stavenga 1979; Land 1997; Rutowski & Warrant 2002). Heads were detached from the body and placed on a goniometer stage under a (a) (b) dissecting microscope (Zeiss Stemi DV4 Spot, Zeiss, Jena, Ger- Hyper- Isometric many). The eye was viewed from the same angle as the incident allometry Isometry microhabitat light using orthodromic illumination (i.e. with the light source effect emanating through the lens). The black area apparent from this view defines the ‘pseudo pupil’ (e.g. Fig. 1e). The ommatidia in line with the view absorb all incident light giving these ommatidia a black appearance. Heads were positioned face upward on the Hypo- goniometer stage at the centre of rotation of one eye so that the allometry pseudopupil moved vertically or medially across the eye. Anatomi- cal references were set for the 0 longitude by centring both pupils of the eyes so that they were equidistant from the centre of the (c) (d) Eye size Eye clypeus. The 0 latitude reference was set to the point where the Non-allometric No relationship centre of the pseudopupil was in line with the top of the clypeus microhabitat with the face perpendicular to the scope (Fig. 1). The eye was effect dusted with chalk powder to allow identification of individual fac- ets across photographs. We obtained a 180 dorsal–ventral tran- sect by photographing the head at 10 steps from positive 90 (top eye) to negative 90 (bottom eye) using a Canon Powershot A650 IS digital camera mounted to the ocular of the microscope. Visual inspection of the eye revealed a single (Movie S1, Supporting information), frontally located acute zone. Thus, we focused on Body size the front dorsoventral transect because this transect encompassed the greatest variation in the traits of interest (Fig. S1). Addition- Fig. 2. Hypotheses for scaling relationships between eye morphol- ally, behavioural observations indicated that male Megalagrion ogy and size. We use eye size as an example, but similar allometric damselflies orient themselves to maximize visual cues in the for- relationships could apply to D or D/. (a) Morphological variables ward direction when perching or flying to explore objects may follow one of several simple scaling relationships with body (R. Schroeder & M. A. Butler, unpublished data), suggesting that size. (b) Alternatively, morphology may be a function of scaling the forward visual field is especially important in these damselflies. and microhabitat effects so that eye morphology scales with size Pseudopupil measurements began with importing the photo- across species, but species cluster by microhabitat. (c) Visual graphs into Image J and drawing a circle around the perimeter of demands of different microhabitats take precedence over size effects the pseudopupil to identify its vertical centre in each photograph. such that species occupying similar microhabitats have similar mor- As the eye is rotated, the pseudopupil moves across the eye. The phology, with distinct differences between microhabitats. (d) No average D/ was measured as 10 (the angular rotation between relationship between visual morphology and size or microhabitat pictures) divided by the number of facets crossed by the centre of use. the pseudopupil along the diagonal frontoventral facet rows.
Table 1. Megalagrion species means, standard errors and sample sizes (N) for body size (thorax length), eye size (eye height) and eye mor- D/ phology variables: maximum facet diameter (Dmax), minimum interommatidial angle ( min) and regionalization in facet diameter (regD), and interommatidial angle (regD/). Regionalization variables are reported as fraction of the eye based on a species mean. Island and breeding microhabitats also given
Body size Eye size ðl D/ ð Þ D/ Species (mm) N (mm) NDmax m) N min N regD reg Island Microhabitat
blackburni 6 25 0 05 25 2 43 0 03 20 32 7 0 98 5 0 627 0 044 5 0 32 0 36 Molokai Stream heterogamias 6 05 0 05 26 2 31 0 02 26 31 4 0 85 5 0 734 0 016 5 0 27 0 41 Kauai Stream oceanicum 5 54 0 11 3 2 06 0 06 3 31 5 0 31 3 0 704 0 048 3 0 26 0 34 Oahu Stream hawaiiense 4 82 0 04 33 1 81 0 01 29 28 8 0 83 5 0 783 0 043 6 0 25 0 28 Molokai Seep koelense 4 56 0 09 7 1 77 0 02 6 28 6 0 50 4 0 782 0 48 4 0 22 0 35 Molokai Plant vagabundum 4 49 0 04 53 1 78 0 01 48 33 8 0 32 6 0 762 0 055 6 0 28 0 27 Kauai Seep oahuense 4 46 0 10 6 1 89 0 04 6 36 7 0 35 4 0 932 0 030 4 0 26 0 37 Oahu Terrestrial n. nigrohamatum 4 39 0 02 21 1 80 0 01 20 35 1 0 31 6 0 574 0 022 6 0 14 0 26 Maui Pool calliphya 4 01 0 03 69 1 54 0 01 54 32 9 0 90 6 0 765 0 045 6 0 18 0 29 Molokai Pool n. nigrolineatum 3 92 0 06 25 1 67 0 01 25 34 2 0 87 6 0 658 0 030 6 0 16 0 27 Oahu Pool xanthomelas 3 89 0 03 32 1 5 0 016 32 34 2 0 42 5 0 823 0 036 5 0 17 0 28 Oahu Pool leptodemas 3 76 0 06 6 1 48 0 03 6 32 3 0 35 6 0 798 0 039 6 0 18 0 27 Oahu Pool oresitrophum 3 43 0 06 34 1 50 0 01 31 32 4 0 49 6 0 830 0 022 6 0 17 0 27 Kauai Pool
© 2015 The Authors. Functional Ecology © 2015 British Ecological Society, Functional Ecology Microhabitat use and damselfly vision 5
Facet diameter measurements were based on the average width We also examined whether eye morphology was explained by of three adjacent facets in a row within each pseudopupil. The scaling or by association with ecological variation using ANOVA. hexagonal geometry of the lens facets results in three possible axes For eye size, we tested models including body size, microhabitat for ommatidial rows (roughly at 60 angles to one another). As type and their interaction. For each functional variable (Dmax, D/ D/ each of these directions results in a similar, but slightly different min, regD and reg ), we tested models with eye size, micro- facet diameter, we took the mean D for each of the three axes habitat type and their interaction as dependent variables, dropping through a central ommatidia and averaged these together. Facet non-significant interactions. All statistical analyses were performed diameters were measured every 5 between 40 and 40 and in the R statistical computing environment (R Core Team 2013). every 10 thereafter to 90 along the same transect as D/ were measured. The raw data were smoothed using a cubic spline in PHYLOGENETIC ANALYSES the R computing environment (R Core Team 2013). The mini- Because species share evolutionary history, they are not indepen- mum interommatidial angle (D/ ) and maximum facet diam- min dent data points for standard statistical analyses (Felsenstein eter (D ) were obtained from the smoothed data. We chose max 1985). To account for phylogeny in our analyses, we used a to analyse minimum D/ and maximum D because they model-based approach to analyse the evolution of quantitative should represent maximum visual abilities. If the visual traits, comparing the fit of Brownian motion and Hansen models. demands of microhabitats are limiting to species distribu- Brownian motion models describe the evolution of traits as a tions, we expect peak abilities at the most sensitive part of the purely stochastic process (Eq. 6; Martins & Hansen 1997; Butler eye to be more closely related to habitat use than average & King 2004). Written as a stochastic differential equation, the abilities. Brownian motion model describes evolutionary change in a phe- notypic trait (X) through time (t) with dB(t) an increment of a EYE REGIONALIZATION standard Brownian motion (BM) process. One parameter is fit in this model, (r) which is proportional to the strength of stochastic As there is no established standard measure of eye regionaliza- changes or macroevolutionary ‘drift’: tion, we used the following index for the degree to which D dXðtÞ¼rdbðtÞ eqn 5 and D/ vary regionally across the eye. We first calculated mean eye transects for each species. We defined the mid-point D size Hansen models are Ornstein–Uhlenbeck processes developed as halfway between the maximum and minimum D obtained for phylogenetic comparative analysis (Eq. 6; Martins & Hansen for each species mean transect (Fig. 5a). Regionalization in D 1997; Butler & King 2004), which describe the evolution of quanti- was therefore the proportion of the transect with facet diame- tative traits under a model of stabilizing selection. It is important ters (regD) above the mid-point D size for the species. Essen- to note that these are used here to represent macroevolutionary tially, regD is the width of the transect at half-max, so that models that describe evolution over very long time-scales (millions highly regionalized eyes will be low in regD. of years) and cannot be extrapolated down to natural selection Using the transects of D as in Fig. 5a, regD, is calculated as within populations, which occur on very short time-scales. Written follows. Let R be a subrange within the degree range (0, 180). as a stochastic differential equation, the Hansen model describes Then, RD is the degree range of the eye with facet diameters that evolutionary change in a phenotypic trait (X) through time (t) are greater than half-max and is defined as follows: with dB(t) an increment of a standard Brownian motion (BM) process, with additional terms for stabilizing selection (a), drift maxðDÞ minðDÞ R ¼ RjD [ eqn 1 (r) and an optimal trait value (h): D 2 dXðtÞ¼a½hðtÞ XðtÞ dt þ rdbðtÞ eqn 6 Then: Importantly, Hansen models allow for the specification of dif- RD regD ¼ eqn 2 ferent optima on different branches of a phylogeny to indicate 180 transitions to different selective regimes, as when species rapidly A similar procedure was used to obtain the index of regionali- evolve in morphology in response to a new microhabitat type. zation in interommatidial angle, regD/: We used the Megalagrion phylogeny (topology and branch lengths) published by Jordan, Simon & Polhemus (2003, Fig. 3). maxðD/Þ minðD/Þ We pruned the Jordan, Simon & Polhemus (2003) tree to include RD/ ¼ RjD/\ eqn 3 2 only the species for which we have phenotypic data using the ‘ape’ and: package in R (Paradis, Claude & Strimmer 2004), and scaled the time depth of the tree to 1. All evolutionary analyses were con- RD/ ducted on log-transformed variables. regD/ ¼ eqn 4 180
Therefore, regD and regD/ can range in value from nearly 0 to Selective regimes nearly 1, with small values indicating greater regionalization. We constructed three hypotheses for the evolution of body size
and eye morphology (Dmax, Dumin, and regionalization of each). STATISTICAL ANALYSES The first was an adaptive model describing adaption to microhabi- tat type. The microhabitat model contains 5 optima based on the Linear regressions of log-eye size with log-body size were microhabitat use of each species, and hypothesizes that at least used to test hypotheses of allometry. Isometry implies that each some of the 5 microhabitat types: stream, seep, plant, terrestrial, linear measurement of morphology will scale one to one with or pool, place different selective pressures on the species that are body length. Therefore, if morphological variation scales isomet- associated with them (Fig. 3). The second model described stabi- rically with size, we expect that a log–log regression will result lizing selection towards a single global optimum, hypothesizing in a slope of 1 0 (Fig. 2a). A slope greater than one indicates that all Megalagrion damselflies experience the same adaptive hyperallometry, and slope < 1 0 indicates hypoallometry. regime. The final model was Brownian motion, which explains
© 2015 The Authors. Functional Ecology © 2015 British Ecological Society, Functional Ecology 6 J.A. Scales & M.A. Butler
heterogamias set was used to create 2000 simulated data sets used for parametric blackburni bootstrap assessment of confidence in model selection and param- oceanicum eter estimation (Burnham & Anderson 2002; Scales, King & Butler kauaiense 2009). Each simulated data set was fit to each model to assess con- koelense fidence in model selection. Models were selected as the best if they hawaiiense performed 2 or more BIC units better than all other models for nesiotes each bootstrap replicate. Confidence intervals for parameter esti- oahuense mates for the best-fitting model were also based on 2000 paramet- mauka ric bootstrap replicates. paludicola vagabundum adytum Results eudytum oresitrophum SIZE AND SCALING calliphya leptodemas Means with standard errors and sample sizes for all mea- orobates surements are given in Table 1. We found that body size in n. nigrohamatum Megalagrion varies over twofold in length and is strongly n. nigrolineatum pacificum associated with microhabitat type (Fig. 4a, Table S1). xanthomelas Stream species are significantly larger than seep (P = 0 01, Tukey HSD test), plant (P = 0 03), terrestrial (P = 0 02) Fig. 3. The Megalagrion phylogeny used in this study, with branch and pool species (P < 0 001). We found support for lengths drawn proportionally to time (Jordan, Simon & Polhemus hypoallometry (i.e. a trend of reduced eye size with increas- 2003). There were thirteen taxa included in this study, indicated ing body size), along with an allometric microhabitat effect by solid lines. Dashed lines indicate extant species not included in this study as some of these are very rare, endangered or may be (Figs 2b and 4b, Table S1). extinct (e.g. M. nesiotes). Breeding microhabitats (Polhemus & Asquith 1996) are mapped onto the phylogeny represented by dif- ferent line colours: red – stream, black – plant, green – seep, light FUNCTIONAL MORPHOLOGY OF THE EYES blue – terrestrial, dark blue – pool. These colour codes are used We found extensive variation in facet diameter, interom- throughout the manuscript. matidial angle and their regionalization, and strong associ- ations with microhabitat type (Fig. 5, Table 2). Despite phenotypic evolution by random drift, making no specific assump- being the smallest in size, pool species have some of tions about selection or adaptation. Each evolutionary model was fit to the data and phylogeny the largest maximum facet diameters (light sensitivity, using the OUCH software package (Butler & King 2004; King & Fig. 5a,c) and smallest interommatidal angle (resolution Butler 2009) in the R statistical computing environment (R Core capabilities, Fig. 5b,d) among Megalagrion damselflies. Team 2013). We performed analyses of power and bias for three Dmax and D/ are able to be simultaneously maximized commonly used information criteria and the Bayesian information min via a high degree of regionalization across the eye so that criteria (BIC) performed the best (Appendix S1). Thus, fits of each D/ model were compared using the BIC, which was used to measure the largest D and smallest occur jointly in a small acute the strength of evidence in support of each competing model zone (Fig. 5a,b). This arrangement has allowed small spe-
(Burnham & Anderson 2002). The best-fitting model for each data cies to have larger Dmax than species which are much larger
(a) (b) Eye size (mm) size Eye Body size (mm) Body size Slope = 0·83 R2 = 0·95 Habitat*** Size*** 3·5 4·0 4·5 5·0 5·5 6·0 1·4 1·6 1·8 2·0 2·2 2·4 Plant Pool Seep Stream Terr 3·5 4·0 4·5 5·0 5·5 6·0 Body size (mm)
Fig. 4. The relationships between microhabitat, body size and eye size in Megalagrion damselflies.(a) Body size was strongly associated with habitat (see Table S1 for ANOVA results). Box plots represent medians with upper and lower quartiles for each ecotype (there are no boxes for terrestrial and plant ecotypes as only one species is represented). (b) Eye size scaled hypoallometrically with body size as a log–log regression of EYEHT on THORAXL revealed a slope of less than one: slope = 0 83 0 063 SE, P < 0.001. Colours and shapes represent microhabitat use in all figures: red circle – stream, black triangle – plant, green square – seep, inverted light blue triangle – terres- trial, dark blue diamond – pool.
© 2015 The Authors. Functional Ecology © 2015 British Ecological Society, Functional Ecology Microhabitat use and damselfly vision 7
Light Sensitivity Variation down eye Max capabilities vs. size Regionalization of eye Terrestrial (a) Terrestrial (c) (e) shady Pool Pool Seep variable shady Seep ) m Stream max Stream reg D bright/large D D ( μ
30 32 34 36 Eye size * Plant 20 25 30 35 Plant Habitat *** Habitat ** bright? Eye size X habitat ** Eye size X habitat * 0·15 0·20 0·25 0·30
(b) (d) (f) Δφ min (degrees) Δφ reg
Δφ Stream bright/large Pool Eye size * Eye size n.s. shady 0·60 0·70 0·80 0·90 Habitat ** 01234 Habitat ** 0·25 0·30 0·35 0·40 −50 0 50 1·6 1·8 2·0 2·2 2·4 1·6 1·8 2·0 2·2 2·4 Eye angle (degrees) Eye size (mm) Eye size (mm) Resolution
Fig. 5. Functional morphology of the eye varies more strongly with microhabitat type than with size in Hawaiian damselflies. The top row displays variation in light sensitivity (D), the bottom row, resolution (D/). Panels (a) D and (b) D/ show examples of variation along a vertical transect down the front of the eye for representative species associated with each microhabitat type (representative species: plant = M. koelense, stream = M. heterogamias, seep = M. vagabundum, pool = M. n. nigrolineatum, terrestrial = M. oahuense; see Figs – D/ – S2 S5 for plots of each species with error bars). From these transects, Dmax and min were obtained for statistical analysis. Panels (c f) show species means with solid lines indicating the fitted regression slope for the pool species (regression analyses were conducted as con- trasts against pool species; i.e. the eye size slope in Table 3). Dotted lines indicate significant interaction effects with eye size (slopes that are significantly different than the main effect slope). Habitat effects (Table 3) can be seen as differences in intercept among the ecomorph D/ categories. (c) Peak light sensitivity (Dmax) varies strongly with habitat, eye size and their interaction, but (d) peak resolution ( min) has weaker effects that lack an interaction. (e) Regional variation in sensitivity (regD) depends on microhabitat and an eye size–habitat inter- action, but (f) regional variation in resolution (regD/) is related to microhabitat and not size (see Table 2 for ANOVA results). *P < 0 05, **P < 0 01, ***P < 0 001.
in size (sharp high peak in D for pool, compared to lower very poor sensitivity (comparatively small Dmax, Table 1, peak for stream Fig. 5a; also compare Dmax in Fig. 5c). Fig. 5a,b). In contrast, the large stream species have less regionalized eyes. While not best in either sensitivity or resolution com- ECOLOGY OR ALLOMETRY pared to other Megalagrion, they had relatively high reso- D/ lution (small D/) and sensitivity (large Dmax) capabilities For three of the four parameters (Dmax, regD and reg ), over more of their eye when compared to species in other ecology explained more variation than allometry among microhabitats (note broader trough in D/ Fig. 5b and in species (compare P-values for the ANOVA effects in D/ Figs S2–S5). Table 2). Only for min was eye size a stronger organiz- Not all microhabitat types maximized both capabilities, ing principle than microhabitat type (eye size effect: as the terrestrial species had the best light sensitivity over- P < 0 02, habitat effect: P < 0 04; Table 2). all (highest Dmax, Fig. 5c) but the worst resolution capabil- D/ ities (largest min, Fig. 5d). In general, resolution Light sensitivity capabilities were more conserved, with greater variation in light sensitivity among microhabitat types. For example, For Dmax, in general, larger eyes had larger Dmax (positive D/ regression slope for eye size Table 3), but there is also evi- the plant and seep species had average resolution ( min) among Megalagrion damselflies, but the plant species had dence for strong ecological effects with highly significant
© 2015 The Authors. Functional Ecology © 2015 British Ecological Society, Functional Ecology 8 J.A. Scales & M.A. Butler
Table 2. ANOVA tables for maximum facet diameter (Dmax), D/ minimum interommatidial angle ( min), regionalization in facet 17 065* 3 diameter (regD) and regionalization in interommatidal angle 0 (regD/). Each variable was tested in separate ANOVAs as dependent variables explained by variation in eye size, microhabitat type and 84 5 236
their interaction as independent variables. Non-significant interac- 0 tions were dropped from the models. P values in bold indicate sig- nificant effects in the model. All models without interactions report type II and models with interaction report type III sums of squares 2** 623 30 Dependent Independent Sum Sq d.f. FP( > F) 0 contained main effects only with no