Biology Letters
You have accessResearch articles

Malar stripe size and prominence in peregrine falcons vary positively with solar radiation: support for the solar glare hypothesis

Michelle Vrettos

Michelle Vrettos

FitzPatrick Institute of African Ornithology, DSI-NRF Centre of Excellence, Department of Biological Sciences, University of Cape Town, South Africa

Google Scholar

Find this author on PubMed

Chevonne Reynolds

Chevonne Reynolds

FitzPatrick Institute of African Ornithology, DSI-NRF Centre of Excellence, Department of Biological Sciences, University of Cape Town, South Africa

School of Animal, Plant and Environmental Sciences, University of the Witwatersrand, Johannesburg, South Africa

Google Scholar

Find this author on PubMed

Arjun Amar

Arjun Amar

FitzPatrick Institute of African Ornithology, DSI-NRF Centre of Excellence, Department of Biological Sciences, University of Cape Town, South Africa

[email protected]

Google Scholar

Find this author on PubMed


    Many falcons (Falco spp.) exhibit a distinct dark plumage patch below the eye, termed the malar stripe. This stripe is hypothesized to reduce the amount of solar glare reflected into the eyes while foraging, thereby increasing hunting efficiency in bright conditions. Here, we use a novel, global-scale correlative approach to test this ‘solar glare hypothesis' in peregrine falcons (Falco peregrinus), the most widespread falcon species, using web-sourced photographs from across the species' global range. We found that the size and prominence of the malar stripe were positively associated with average annual solar radiation, but not with other environmental variables, such as temperature and rainfall. Our results provide the first published evidence for the hypothesis that this plumage feature functions to reduce the amount of solar glare reflected into the falcon's eyes, thereby improving the ability to pinpoint and target agile prey in bright conditions.

    1. Introduction

    Animal markings may serve a variety of functions, including crypsis, communication and thermoregulation [14]. They can thus be affected by multiple, often opposing selective forces, typically representing trade-offs between different biotic and abiotic pressures [46]. Local conditions can affect the relative importance of these selective forces, potentially resulting in regional colour variation [4,710]. However, identifying the evolutionary pressures that drive regional variation is challenging because of trade-offs involved in the evolution and maintenance of coloration patterns [4].

    Dark eye markings are found in many animal taxa [2,11,12]. They typically occur in species associated with bright habitats [2,11] and have therefore been hypothesized to function by reducing the amount of solar glare reflected into the eyes while foraging, thereby improving the ability to distinguish and target food objects in bright conditions [13,14]. This ‘solar glare hypothesis' for the function of dark eye markings is supported by experimental evidence from both birds [1,13,15,16] and humans [17]. Experimental lightening of the upper mandibles of willow flycatchers (Empidonax traillii) and facial masks of masked shrikes (Lanius nubicus) were associated with reduced foraging efficiency in sunlight and accordant shifts in foraging behaviour towards more shaded locations [13,14], while wood-warbler (Parulinae) species with dark upper mandibles reportedly forage more frequently in sunlight than do those with light-coloured upper mandibles [1]. Dark facial plumage has also been shown to reduce the amount of irradiance reflected into the eye across multiple bird species [16], while ‘eye black’ grease worn by athletes reportedly improves human subjects' contrast sensitivity and ability to see detail in bright conditions [17].

    Many falcons (Falco spp.) exhibit a distinctive stripe of dark feathers below the eye, termed the malar stripe. The solar glare hypothesis, as described above, is often invoked to explain the function of this plumage feature [3,18], but has never been quantitatively tested. Indeed, some have criticized this hypothesis, since some falcon species that inhabit very bright environments (e.g. deserts) have relatively pale malar stripes, or lack them completely [3]. Peregrine falcons (Falco peregrinus) (hereafter peregrines) are among the most widely distributed of all bird species [19,20] and exhibit considerable geographic variation in malar stripe characteristics [2024]. Solar glare impedes vision by introducing non-information-containing light onto the retinal image, thereby reducing the contrast of the image and ability to see detail or resolve objects from the background [17]. This reduced contrast sensitivity hinders the subject's ability to distinguish and target moving objects as their relative velocity increases or decreases [17]. Peregrines are diurnal, aerial predators that mainly hunt agile birds [19,24,25]. By absorbing some of the excess light that causes glare and counteracting this effect on contrast sensitivity, the malar stripe may thus improve peregrines' hunting efficiency in bright conditions by increasing their ability to distinguish and pinpoint fast-moving prey. While birds’ contrast sensitivity is lower than that of mammals (and that of peregrines is unknown), American kestrels (Falco sparverius) have the highest contrast sensitivity of all birds tested [26], while other features of falcon eye anatomy (e.g. supraorbital ridges, optical adnexa and pecten oculi) are also hypothesized to function as glare-reduction devices, suggesting that features that reduce the impact of solar glare on vision are strongly selected for in falcons [27,28]. If peregrine malar stripes function to increase hunting efficiency in bright conditions, as proposed by the solar glare hypothesis, peregrines may be expected to exhibit larger or darker malar stripes in regions of higher average solar radiation.

    Other plausible adaptive hypotheses for falcon malar stripes include those associated with Gloger's and Bogert's ecogeographical rules. Gloger's rule predicts that darker coloured individuals should occur in wetter areas [4,9,2931]. This pattern may arise due to selection against higher parasite loads in wetter areas, either through melanization increasing resistance to ectoparasite damage [32,33], or through immune advantages linked pleiotropically to darker plumage [7,3437]. Alternately, darker individuals may experience better crypsis in the poor light conditions associated with wetter areas [4,10,38,39]. If peregrine malar stripes served either of these adaptive functions (or simply covaried with overall plumage trends, which follow Gloger's rule [40]), peregrines would be predicted to exhibit larger malar stripes in regions of higher rainfall. Alternatively, the thermal melanism hypothesis, or Bogert's rule, predicts that animals in colder environments should be darker than those in warmer environments, due to thermal advantages conferred by dark pigmentation [4,29,41,42]. Darker pigmentation increases heat absorption and lowers thermal reflectivity, thereby reducing the need to increase metabolism at low temperatures [30,4346]. If this hypothesis explained the adaptive function of dark facial plumage in falcons, or if facial plumage reflected a trend towards overall darker plumage in colder areas, we would predict larger malar stripes in peregrines inhabiting colder regions.

    The pan-global distribution of peregrines means that they occupy terrestrial habitats spanning a wide range of environmental conditions, making this species ideally suited for testing adaptive hypotheses for the function of falcon malar stripes. In this correlative study, we use a unique, web- and citizen science-based approach [47] to obtain photographs from across the species' range, in order to quantify global-scale variation in malar stripe characteristics. We then relate this to variation in local climatic conditions to determine which of the aforementioned adaptive hypotheses for the function of peregrine malar stripes is best supported by observed ecogeographical trends.

    2. Methods

    (a) Sourcing photographs

    We obtained photographs of adult peregrines from two online citizen science repositories, the Macaulay Library ( and iNaturalist ( We obtained a maximum of 50 photographs for each country, except the USA and Canada, which cover large geographic areas and for which there were many photographs available. For these, we instead obtained a maximum of 50 photographs per-state or province, treating these as ‘countries’. See the electronic supplementary material for full details regarding how photographs were selected.

    (b) Scoring photographs

    We quantified five malar stripe characters: (i) width, (ii) contiguity with the hood, (iii) prominence and (iv) length (detailed definitions provided in the electronic supplementary material, figures S1 and S2), along with (v) an ‘elongation’ measure, calculated by dividing length score by width score. A single observer (M.V.) visually scored each character on a 10-point scale (electronic supplementary material, figure S2). Photographs were scored in random order with the observer unaware of their location. The angle and distortion of the head were also scored on a visual scale (electronic supplementary material, figure S3), since we suspected that the bird's posture could potentially influence the malar stripe scores.

    (c) Climate variables

    For each photograph location, we extracted average minimum daily temperatures (°C), annual rainfall (mm) and annual solar radiation (W m−2) from the TerraClimate dataset, which provides climate data for global terrestrial surfaces at an approximately 4 km2 spatial resolution [48]. For each climate variable, we extracted 30-year (1988–2018) averages within a 20 km buffer of the photograph's GPS location, or for the general area, for cases in which GPS coordinates were unavailable (N = 17).

    (d) Statistical analysis

    We explored relationships between the five malar stripe characters and three climate variables using generalized least-squares (GLS) models, using the R package nlme [49]. We incorporated the GPS coordinates of each photograph, using an exponential spatial autocorrelation structure, to account for any spatial autocorrelation in the data [50].

    For each malar stripe (response) variable, we built nine candidate GLS models, including a null model, a model containing only the head position variables (electronic supplementary material, figure S3) and seven models containing these plus all possible combinations of the three climate variables. All explanatory variables were standardized to aid interpretation and allow direct comparisons of their effects [51]. While the malar stripe has been reported to be sexually dimorphic in peregrines [20], we were unable to test or control for a sex effect, as most birds in our photographs were unsexed.

    For each response variable, candidate models were compared using their AICc values. Plausible models were defined as those within 2 AICc units of the top-ranked model. Model-averaged parameter estimates (±95% CI) for the three climate variables were then extracted using the AICcmodavg package [52].

    (e) Validation of scores and shape analysis

    To validate the subjective scores, and to investigate further variation in malar stripe shape in our dataset, we selected a subset of observations of similar photograph quality and angle (N = 213). For this subset, we calculated quantitative measurements of malar stripe size, shape and prominence (grey value) using ImageJ [53]. Full details of these analyses are provided in the electronic supplementary material.

    3. Results

    We analysed images depicting a total of 2197 peregrines from 96 ‘countries’ (electronic supplementary material, table S1). We found strong statistical support for positive relationships between annual solar radiation and four of the five malar stripe variables (table 1 and figure 1). Peregrine malar stripes were wider, more contiguous with the hood, more prominent (darker) and less elongated (wider relative to length) at locations experiencing higher average annual solar radiation (figures 1 and 2; electronic supplementary material, figure S6).

    Table 1. Model selection table showing plausible candidate GLS linear regression models for each of the five malar stripe variables, ranked based on AICc values. Models were considered plausible if they fell within 2 AICc units of the top-ranked model.

    variable rank explanatory variables residual standard error residual d.f. log likelihood AICc delta AICc AICc weight
    width 1 solar radiation + min. temp 1.4577 2191 –3945.15 7906.37 0.00 0.67
    2 solar radiation + rainfall + min. temp 1.4575 2190 −3944.87 7907.82 1.45 0.33
    contiguity 1 solar radiation + rainfall 2.1734 2191 −4822.90 9661.87 0.00 0.62
    2 solar radiation + rainfall + min. temp 2.1733 2190 −4822.80 9663.68 1.81 0.25
    prominence 1 solar radiation 1.2759 2192 −3638.18 7290.40 0.00 0.41
    2 solar radiation + min. temp 1.2758 2191 −3637.92 7291.91 1.50 0.19
    3 solar radiation + rainfall 1.2758 2191 −3637.93 7291.93 1.53 0.19
    4 solar radiation + rainfall + min. temp 1.2752 2190 −3636.94 7291.96 1.56 0.19
    length 1 head position variables only 1.1300 2193 −3385.97 6783.98 0.00 0.26
    2 min. temp 1.1297 2192 −3385.42 6784.90 0.91 0.16
    3 solar radiation + rainfall + min. temp 1.1288 2190 −3383.58 6785.25 1.27 0.14
    4 solar radiation + min. temp 1.1294 2191 −3384.79 6785.64 1.66 0.11
    5 rainfall 1.1300 2192 −3385.91 6785.88 1.90 0.10
    6 solar radiation 1.1300 2192 −3385.96 6785.98 2.00 0.09
    elongation (length/width) 1 solar radiation + min. temp 0.3404 2191 −749.76 1515.59 0.00 0.53
    2 solar radiation + rainfall + min. temp 0.3403 2190 −748.90 1515.87 0.28 0.46
    Figure 1.

    Figure 1. Model-averaged parameter estimates for the effect of average annual solar radiation (W m−2), average annual rainfall (mm) and average minimum daily temperature (°C) on each of the five malar stripe variables, with bands representing 95% confidence intervals. Explanatory variables for which the 95% confidence interval overlaps zero (represented by the vertical dotted line) are considered to have no effect on the response. Parameter estimates for malar stripe elongation are plotted on a separate axis to enable easier comparison of effect sizes, since this variable was not scored on a 10-point scale. Plots were constructed using the R package ggplot2 [54].

    Figure 2.

    Figure 2. Maps showing mean malar stripe width scores for individual peregrines in web-sourced photographs, both (a) per country (N = 2197) and (b) per state for the USA and Canada (N = 1433), as a function of average annual solar radiation (W m−2). Mean malar stripe width scores for each country/state are represented by bubble size, while bubble colour indicates sample size for each country/state. Maps were constructed using the R package rworldmap [55].

    Minimum temperature was associated only with malar stripe width and elongation (table 1 and figure 1); peregrines had wider, less elongated malar stripes in colder areas (figure 1; electronic supplementary material, figure S6). Similarly, rainfall was associated only with contiguity (table 1 and figure 1), with malar stripes less contiguous with the hood in wetter areas (figure 1; electronic supplementary material, figure S6).

    For a subset of the data (N = 213), we found significant (p < 0.001) correlations between the malar stripe width, contiguity and length scores and quantitative measurements of malar stripe area; between prominence scores and grey value measurements; and between elongation scores and quantitative shape measurements (electronic supplementary material, figure S5).

    4. Discussion

    The solar glare hypothesis for the function of falcon malar stripes has become widely accepted in popular literature, despite the paucity of studies on the topic [3]. Our study represents the first quantitative investigation of this hypothesis. We found strong statistical support for the predictions of the solar glare hypothesis; peregrines with wider and more prominent malar stripes or overall darker heads were associated with areas of higher annual solar radiation. This pattern matches those found in passerine birds and mammalian carnivores, in which dark eye markings are associated with species which forage in bright environments [2,11]. Bortolotti [3] argued that falcon malar stripes were unlikely to serve a glare-reducing function, since desert-dwelling saker (Falco cherrug) and prairie falcons (F. mexicanus) have pale malar stripes, while forest-dwelling orange-breasted (F. deiroleucus) and bat falcons (F. rufigularis) have almost completely black heads. However, this argument did not consider how phylogenetic signal and differences in diet and body size may affect the relative adaptive or functional importance of malar stripes between species. Saker and prairie falcons, unlike peregrines, often prey on mammals and typically catch their prey on the ground, while orange-breasted and bat falcons predominantly hunt aerial prey [19], and may thus, like peregrines, potentially benefit from adaptations that increase their ability to pinpoint fast-moving targets while exposed to bright sunlight. Indeed, the fact that diet and body size also vary intraspecifically in peregrines [40,56] may account for some of the unexplained variation in our data. Our findings are also consistent with the empirically demonstrated roles of dark under-eye markings in both birds [1,13,15,16] and humans [17].

    We found no evidence that peregrine malar stripe characteristics follow Gloger's rule. Indeed, the only malar stripe character associated with rainfall (contiguity) exhibited a pattern opposite to that expected under Gloger's rule, with peregrines with darker heads associated with drier regions. This suggests that the malar stripe is unlikely to serve a cryptic or background-matching function in peregrines [10,39]. This is perhaps unsurprising, given that the malar stripe covers only a small area of the body and is presumably less conspicuous to prey than other ventral plumage, and thus unlikely to be important in camouflage. Peregrines have been reported to follow Gloger's rule in terms of overall body plumage [40]. Our results thus suggest that malar stripe characteristics are not correlated with overall plumage darkness in peregrines, and that the relationships we observed are not due to geographic trends in overall body plumage.

    We found significant relationships with temperature for only two malar stripe variables, width and elongation. The directions of these relationships were consistent with the predictions of the thermal melanism hypothesis, with peregrines with wider, less elongated (blockier) malar stripes associated with colder regions. However, the fact that the other malar stripe measurements did not vary consistently with temperature suggests that peregrine malar stripes do not generally follow Bogert's rule, in contrast with patterns observed in plumage coloration in other raptors [4,31] and other bird groups [30,44]. Thus, malar plumage does not appear to play a significant thermoregulatory role in peregrines.

    While this study provides some evidence in support of the solar glare hypothesis for the function of peregrine malar stripes, our analysis nevertheless suffers from several weaknesses. In addition to our analysis relying purely on statistical correlation, the effect sizes we observed were relatively small, and there was considerable unexplained variation in the data. Thus, while the trends we observed were statistically significant, their biological relevance is unclear and potentially limited [57]. Our study also relies on subjective measurements, which may be vulnerable to observer effects. However, a similar study [8] found that subjective scores of plumage characteristics from multiple independent observers did not differ significantly, which suggests both that this method is robust to observer effects, and that a single observer is sufficient to produce reproducible results. Our subjective scores also correlated significantly with quantitative measurements of malar stripe size, shape and prominence for a representative subset of the data (electronic supplementary material, figure S5).

    Additionally, because photographs were captured throughout the year and across the species' entire range, our dataset likely included birds that were wintering or passage migrants, while many from eastern North America were likely non-native subspecies hybrids, owing to the restoration of the species in this region from captive individuals [58]. To check whether these issues could have influenced the results, we performed two repeat analyses. In the first, we included only photographs depicting non-migratory subspecies and/or taken within the northern hemisphere breeding season (March–August), while in the second, we excluded all photographs from the USA and Canada. Most of the statistical relationships were not meaningfully altered in these new analyses (electronic supplementary material, tables S3 and S4).

    In conclusion, our results suggest that the function of the malar stripe in peregrines is best explained by the solar glare hypothesis. The malar stripe is therefore more likely to serve a solar glare-reducing function than it is a cryptic or thermoregulatory function. However, the evidence we present for this conclusion is purely correlative, and future research could test the predictions of the solar glare hypothesis experimentally. Future studies could likewise extend this analysis across the genus Falco, both to elucidate the functional and evolutionary significance of malar stripes in falcons generally, and to investigate how differences in feeding ecology may affect the function and relative adaptive importance of malar stripes in different falcon species.


    No ethical approval was required for this study.

    Data accessibility

    All data and code for this study are available at the following Dryad link: [59].

    The data are provided in the electronic supplementary material [59].

    Authors' contributions

    A.A. conceived of the idea, coordinated the study, oversaw the data collection and analysis, helped draft the manuscript and provided critical revisions to the manuscript. M.V. performed the data collection, data processing and statistical analysis, produced all visuals and figures, and drafted and revised the manuscript. C.R. extracted the climatic data for the analysis, helped draft the manuscript and provided critical revisions to the manuscript. All authors gave final approval for publication and agree to be held accountable for the work performed therein.

    Competing interests

    We declare we have no competing interests


    This study was supported by National Research Foundation (grant no. 40470).


    We are grateful to the staff and curators of the Macaulay Library at the Cornell Lab of Ornithology and the iNaturalist database, and to the ‘citizen scientists’ who contributed their images. A list of all images used is provided in the Appendices, included here as supplementary material. We also thank the Handling Editor and the anonymous reviewers for their insightful comments.


    Second e-mail:

    Electronic supplementary material is available online at

    Published by the Royal Society. All rights reserved.