OP-BIOM170003 1..8 Q u a n t ifyi n g p i g m e n t c o v e r t o a s s e s s v a r i a t i o n i n a n i m a l c ol o u r a t i o n S i e g e n t h a l e r, A, M o n d a l , D a n d B e n v e n u t o , C h t t p :// d x . d o i. o r g / 1 0 . 1 0 9 3 / b i o m e t h o d s / b p x 0 0 3 T i t l e Q u a n t ifyi n g p i g m e n t c o v e r t o a s s e s s v a r i a t i o n i n a n i m a l c ol o u r a t i o n A u t h o r s S i e g e n t h a l e r, A, M o n d a l , D a n d B e n v e n u t o , C Ty p e Ar ti cl e U R L T hi s v e r s i o n i s a v a il a b l e a t : h t t p : // u s ir. s a lf o r d . a c . u k /i d / e p r i n t / 4 1 5 4 4 / P u b l i s h e d D a t e 2 0 1 7 U S I R i s a d i gi t a l c oll e c t i o n of t h e r e s e a r c h o u t p u t of t h e U n iv e r s i t y of S a lf o r d . W h e r e c o p y r i g h t p e r m i t s , f ull t e x t m a t e r i a l h e l d i n t h e r e p o s i t o r y i s m a d e f r e e l y a v a il a b l e o n li n e a n d c a n b e r e a d , d o w n l o a d e d a n d c o p i e d fo r n o n- c o m m e r c i a l p r i v a t e s t u d y o r r e s e a r c h p u r p o s e s . P l e a s e c h e c k t h e m a n u s c r i p t fo r a n y f u r t h e r c o p y r i g h t r e s t r i c ti o n s . F o r m o r e i nf o r m a t i o n , i n cl u d i n g o u r p o li c y a n d s u b m i s s i o n p r o c e d u r e , p l e a s e c o n t a c t t h e R e p o s i t o r y Te a m a t : u s i r @ s a lf o r d . a c . u k . mailto:usir@salford.ac.uk M E T H O D S M A N U S C R I P T Quantifying pigment cover to assess variation in animal colouration Andjin Siegenthaler, Debapriya Mondal, and Chiara Benvenuto* School of Environment and Life Sciences, University of Salford, Salford M5 4WT, UK *Correspondence address. School of Environment and Life Sciences, Room 317, Peel Building, University of Salford, Salford M5 4WT, UK. Tel: þ44 (0)161- 295-5141; E-mail: C.Benvenuto@salford.ac.uk Abstract The study of animal colouration addresses fundamental and applied aspects relevant to a wide range of fields, including behavioural ecology, environmental adaptation and visual ecology. Although a variety of methods are available to measure animal colours, only few focus on chromatophores (specialized cells containing pigments) and pigment migration. Here, we illustrate a freely available and user-friendly method to quantify pigment cover (PiC) with high precision and low effort us- ing digital images, where the foreground (i.e. pigments in chromatophores) can be detected and separated from the back- ground. Images of the brown shrimp, Crangon crangon, were used to compare PiC with the traditional Chromatophore Index (CI). Results indicate that PiC outcompetes CI for pigment detection and transparency measures in terms of speed, accuracy and precision. The proposed methodology provides researchers with a useful tool to answer essential physiological, behav- ioural and evolutionary questions on animal colouration in a wide range of species. Keywords: chromatophores; chromatosomes; Chromatophore Index; colour change; colour threshold; Crangon crangon; ImageJ Introduction The study of animal colouration and colour patterns is essential to gather a better understanding on how animals visually com- municate and how they can match different substrates. Furthermore, this type of studies provides important insights on how predation avoidance due to camouflage can drive inter- and intraspecific variation, and how colouration and visual per- ception are connected (e.g. [1]). A wide range of methods has been developed to measure animal colouration, which can be roughly divided in three categories: (i) spectral quantification of colouration and animal vision [2, 3]; (ii) assessment of colour patterns [4–7]; and (iii) analysis of chromatophores and pigment migration [8–10]. The last method has been used mainly to study animal colour changes [9, 11, 12]. Chromatophores are specialized cells containing pigmented organelles and can be located in the dermis, epidermis, beneath a translucent exoskeleton, deep in muscular tissue or around internal organs [13–15]. In crustaceans, multiple tightly bound chromatophores (of similar or different colours) are combined in a structure called chromatosome [13, 16]. Many animals can regulate their colour by the dispersal and concentration of pig- ments within chromatophores (e.g. [12, 17]): colour can be changed in a period of days to months through anabolism and catabolism of pigments and cells (morphological colour change) or within milliseconds to hours via the migration of pigments within chromatophores (physiological colour change) [12]. The concentration or dispersion of pigments reduces or increases their visibility, since less or more surface area is covered by them, respectively [18, 19]. Hogben and Slome [20] described changes in the pigment distribution in the frog Xenopus laevis by classifying chromatophores in five classes (Supplementary Fig. S1), applying a Melanophore Index (MI) for melanophores (also more generally called Chromatophore Index (CI) for chromato- phores containing pigments other than melanin [20, 21]). Although this method has been extensively used (see Table 1 for some recent examples), concerns have been raised about its Received: 27 September 2016; Revised: 15 February 2017; Editorial decision: 17 February 2017; Accepted: 2 March 2017 VC The Author 2017. Published by Oxford University Press. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. 1 Biology Methods and Protocols, 2017, 1–8 doi: 10.1093/biomethods/bpx003 Methods Manuscript Deleted Text: INTRODUCTION Deleted Text: - Deleted Text: [ Deleted Text: 1 Deleted Text: 2 Deleted Text: 3 Deleted Text: s Deleted Text: [ Deleted Text: 5 http://www.oxfordjournals.org/ degree of subjectivity, statistical validity and labour intensive- ness [22, 23]. Here, we describe a new method, PiC (Pigment Cover), to assess the degree of pigment dispersion within chro- matophores (or chromatosomes) by measuring the coverage of pigments in defined areas of an animal body, thus allowing us to evaluate colour variations in a quantitative way. The objec- tive of this study is to demonstrate the use and versatility of PiC and compare it to the established CI. To achieve this, both PiC and CI were applied to a database of pictures of the brown shrimp, Crangon crangon (L.), a crustacean characterized by good background-matching abilities [8]. Material and methods Protocol to measure PiC Image acquisition Measurements on animal colour or pigment migration are usu- ally performed on a specific body region rather than the whole animal [1, 10, 39]. In some cases, e.g. fish scales [36], the area of interest can be separated from the animal prior to image acqui- sition, reducing the effects of animal stress on the colour [40]. The specimen should be placed and photographed on a uniform surface (Fig. 1A). Contrast between background and pigments should be as high as possible; overlap with underlying organs should be avoided, if possible [23]. The magnification should be high enough to distinguish individual chromatosomes. If multi- ple pigments are studied, the collection of multiple images of the same area on different backgrounds might be necessary (see below). To optimize image acquisition, illumination within an image should be uniform and shadows or reflection of light should be avoided. Light conditions are, nevertheless, less con- stricted than in other methods (e.g. [3, 41]) and colour charts are not required (they can vary in quality and applicability; [2, 3]). Still, standardization of lighting conditions and camera settings will significantly reduce the use of manual adaptations during image analysis (see [41] for more information on the standar- dization of digital images). In digital photography, images are commonly displayed in a non-linear standard default colour space (sRGB). PiC can be applied to these standard images. For more rigorous and objective image analyses, linear images are often required. If this is the case, sRGB images can be converted to the CIELAB colour space using the ‘Color Space Converter’ plugin of ImageJ (https://imagej.nih.gov/ij/plugins/color-space- converter.html). A normalization step is advised to slightly en- hance the contrast within the images by using the ‘Enhance Contrast’ command of ImageJ. A slight over-saturation of 1% is advised for improved visual evaluation [48]. Colour threshold PiC image analysis can be performed with any graphic editor able to perform image segmentation (partitioning an image into sets of pixels) by means of thresholding. Image segmentation by semi-automatic thresholding is an established method that has been used in a range of biological studies, including crop root length [42], plant signals [43] and cell counts [44], but not specifically on pigment coverage. The methodology described in this section is tailored to the freely available java-based im- aging program ImageJ (1.48v, http://imagej.nih.gov/ij/; [45]; RRID:SCR_003070) because of its ease of use and efficacy, but could easily be adapted to other graphic software. Images need to be cropped to the region of interest and seg- mented to differentiate foreground (the pigments under study) and background [46]. In ImageJ, sRGB image segmentation is achieved with the ‘Color Threshold’ function (Fig. 1B), which seg- ments 24-bit RGB images based on pixel values (see the ImageJ user guide; [47]). A range of automatic thresholding algorithms is available in ImageJ. These algorithms perform differently depend- ing on the distribution of pixel values in the image and the most suitable thresholding algorithm should be selected prior to analysis [42, 48], e.g. using the ‘Threshold Check’ macro of the BioVoxxel toolbox (http://www.biovoxxel.de/development/, http:// fiji.sc/BioVoxxel_Toolbox#Threshold_Check). The sensitivity of the threshold function can be manually adapted using the ‘Saturation’ and ‘Brightness’ scroll bar in the colour threshold set- tings window (Fig. 1B) until the whole area covered by the pig- ment(s) of interest is selected [44, 49]. Manual alteration of the thresholding level reduces, however, the objectivity of the analysis and should be avoided as much as possible. Specific pigments can also be selected by adapting the ‘Hue’ scroll bar (Fig. 1B) to the Table 1: Selected publications applying the MI of Hogben and Slome [20] Group Species Area of interest Topic Method Source Amphibian Bufo melanostictus Dorsal skin Drug development MIa [24] Hoplobatrachus tigerinus Isolated dorsal skin cell Physiology MIa [25] Rana catesbeiana Dorsal skin Endocrinology MI [26] Taricha granulosa Larva UV protection MI [27] Ambystoma gracile Larva UV protection MI [27] Ambystoma macrodactylum Larva UV protection MI [27] Xenopus laevis Larva Developmental biology MI [28, 29] Crustacean Chasmagnathus granulata Maxilliped’s meropodit UV protection CI [30] Palaemonetes argentinus Dorsal abdomen UV protection CI [30] Eurydice pulchra Not specified Endocrinology CIa [31] Palaemon pacificus Dorsal abdomen Endocrinology CI [32, 33] Reptile Hemidactylus flaviviridis Dorsal skin Drug development MIa [34] Teleost Ctenopharyngodon idellus Scale Physiology MI [35] Danio rerio Scale and embryo Physiology MI [36] Oncorhynchus mykiss Scale Ecotoxicology MI [37] Verasper moseri Base of caudal fin Developmental biology MI [38] aModified index; MI, Melanophore Index (pigment is melanin); CI, Chromatophore Index (pigment is not melanin). 2 | Siegenthaler et al. Deleted Text: MATERIAL AND METHODS Deleted Text: pigment cover Deleted Text: PROTOCOL TO MEASURE PIGMENT COVER Deleted Text: , Deleted Text: [ Deleted Text: [ Deleted Text: s Deleted Text: s Deleted Text: `` Deleted Text: '' https://imagej.nih.gov/ij/plugins/color-space-converter.html https://imagej.nih.gov/ij/plugins/color-space-converter.html Deleted Text: s Deleted Text: Pigment cover Deleted Text: which Deleted Text: [ http://imagej.nih.gov/ij/ Deleted Text: [ Deleted Text: , http://www.biovoxxel.de/development/ http://fiji.sc/BioVoxxel_Toolbox#Threshold_Check http://fiji.sc/BioVoxxel_Toolbox#Threshold_Check required hue values [43]. For transparency measurements, the ‘Hue’ scroll bar should be used to select the background colour to ensure that only the transparent area is selected (the background will be visible through the transparent tissue) and all pigments are ignored. In cases where only one channel of the image is analysed (e.g. CIELAB’s L channel or greyscale images), ImageJ’s ‘Threshold’ function can be used in similar fashion as the ‘Color Threshold’ function. PiC analysis The area of the selected pigment(s) can be calculated with the ‘Analyze Particles’ command (Fig. 1C) that measures ‘particles’ (separate shaped objects) in an image after thresholding by scanning the image and outlining the edge of objects found has been performed [47, 50]. Case study Dark and light pigment measurements and transparency Five specimens of C. crangon were selected based on visual dif- ferences in colour. Their right exopod (the external branch of their tail fan) was photographed under a stereo microscope (Leica S6D) with a Leica DFC295 camera. The tail fan is the most suitable body area of caridean shrimp to be used for monitoring chromatic parameters because: (i) it is very flat; (ii) it has no un- derlying organs or tissue (and is thus highly transparent); and (iii) it can be photographed while causing minimal stress to the animal [23, 51]. Artificial illumination was provided by two led spotlights (JANSJ €O; 88 lm; 3000 Kelvin) positioned at either side of the microscope. We adjusted the white balance prior to im- age collection and allowed the exposure time to be automati- cally adapted. Images were collected in sequence, on four differently coloured backgrounds (Fig. 2): white for the measurement of dark-coloured (black and sepia-brown) pig- ments; black for light-coloured (white and yellow) pigments and green and blue for transparency measurements. Green and blue hues do not occur naturally in C. crangon [8, 51] and are, there- fore, suitable for transparency measurements (both colours were used in order to test which one performs better). To avoid adaptation to the background during the measurements, shrimp were kept for a very short duration only (less than 1 min) on each background. Images were saved in uncompressed TIFF format (RGB), cropped to 1 mm2 and analysed following the protocol described above, using the default thresholding method, based on the IsoData algorithm [52, 53], and manually adapted if needed. We selected the default thresholding algo- rithm for this experiment since it performed best for the variety of features (dark pigment, light pigment, transparency) tested. For the same photos, we determined the CI, in accordance to the method of Hogben and Slome [20], by classifying all chroma- tosomes in the selected area individually and averaging their values (see Supplementary Fig. S1 for reference). Dark PiC and CI comparison Fifty sRGB images of C. crangon (Fig. 3; obtained from 36 individual shrimp) were selected to represent the range of colouration shown by shrimp (lighter or darker, depending on the substrate where animals were kept). We tested the robustness of the meth- odology used by selecting images varying in properties such as il- lumination and picture quality. All images were obtained on a white background and cropped to 1 mm2 in the centre of the exo- pod. Images were analysed for dark pigments, which are the most abundant and evident pigments responsible for dark col- ouration [8, 51]. Three observers analysed the images with the PiC and CI methods, in random order. Prior to analysis, we ap- plied a threshold check (BioVoxxel toolbox) to a sub-selection of Figure 1: Protocol for PiC measurements. This diagram outlines the steps to be performed in ImageJ to determine PiC. See text for details. Animal pigment cover quantification | 3 Deleted Text: a Deleted Text: Pigment cover Deleted Text: which Deleted Text: `` Deleted Text: '' Deleted Text: CASE STUDY Deleted Text: 1 Deleted Text: , Deleted Text: 2 Deleted Text: 3 Deleted Text: one Deleted Text: ute Deleted Text: and Deleted Text: P Deleted Text: p Deleted Text: igment C Deleted Text: c Deleted Text: over Deleted Text: C Deleted Text: c Deleted Text: hromatophore I Deleted Text: i Deleted Text: ndex 13 images to determine the optimal thresholding algorithm. Based on the average score of these images, the MaxEntropy al- gorithm [54] was selected for all images. Manual adaptation was applied as little as possible (on average on 23% of the images, de- pending on the observer). To test the effect of image linearization and normalization, the 50 sRGB images were transformed to the CIELAB colour space and the L channel was normalized prior to PiC determination. The MaxEntropy thresholding algorithm was applied and, in this case, no manual adaptation was allowed to eliminate the need for subjective input. PiC values of the sRGB (averaged over the observers) and linearized/normalized images were compared using linear regression. Data analyses Inter-observer variation for both dark PiC and CI was tested with the Friedman’s test. This statistical test was selected be- cause of the non-normal distributed nature of both proportions and ordinal data, and the fact that each image was tested re- peatedly. Both PiC (percentage of PiC transformed to fraction) and CI results were averaged between observers and a beta re- gression (betareg R package; [55]) was used to compare the methods. This specific analysis can also be important to predict the results from one method (PiC) when having information from the other (CI). Different link functions (log, log–log and logit) were compared based on Akaike Information Criterion (AIC). Beta regression is considered a suitable test for non- parametric and bounded data such as proportions [55]. Data analyses were performed with R statistical software v.3.1.2 (RRID:SCR_001905) and IBM SPSS statistics v. 20 (RRID:SCR_002865). Results Dark and light pigment measurements and transparency For the five specimens analysed, dark PiC values ranged from 8.9% to 92.1% and light PiC values from 0.8% to 10.6% (Fig. 2). Transparency measurement ranged from 19.8% to 89.2% on a blue background and from 18.1% to 84.3% on a green back- ground (mean difference 6 SD: 1 6 5.9%) and did not significantly differ between the background colours (Wilcoxon signed-ranks test: N¼5, Z¼�0.674, P¼0.500). CI, by definition, cannot be calculated for transparency (Fig. 2). When dark pig- ments were predominant (e.g. shrimps 4 and 5 in Fig. 2), the CI of light pigments could not be calculated, as it was impossible to distinguish the chromatosomes’ shape. Furthermore, the high overlap of dark chromatosomes made it impossible to count the number of chromatosomes to calculate the mean dark CI. In these cases, the CI was estimated as 5, the maximum index value. Figure 2: Pigment and transparency values for cover (%) and CI for five shrimps (C. crangon) on different backgrounds. For each specimen, the right exopod was photo- graphed, always in the same exact position and then the image cropped in the centre (selecting 1 mm2). Red areas represent the area selected by the PiC method. NA: CI cannot be calculated. *CI is an estimate. 4 | Siegenthaler et al. Deleted Text: s Deleted Text: s Deleted Text: s Deleted Text: DATA ANALYSES Deleted Text: pigment cover Deleted Text: pigment cover Deleted Text: [ Deleted Text: - Deleted Text: RESULTS Deleted Text: . Deleted Text: . Deleted Text: S Deleted Text: R Deleted Text: T Deleted Text: - Deleted Text: , Figure 3: Percentage dark PiC and CI of 50 images (1 mm2) of C. crangon’s exopods. The images show different levels of chromatosome dispersion and represent the range of colouration exhibited by the animals. See text for information on capital letters. Animal pigment cover quantification | 5 Dark PiC and CI comparison PiC and CI for all 50 images were calculated (Fig. 3). Dark PiC showed a strong exponential relationship with CI (Fig. 4) and the beta regression confirmed a significant relationship be- tween PiC and CI (coefficient 6 SEM: 0.659 6 0.034; P < 0.0001) with a pseudo-R2 value of 0.95. The equation to estimate PiC from a known CI value was modelled as: Ln ðpredicted PiCÞ¼�3:362 þ 0:659 � CI The equation is only valid for: 1�CI�5 and 0�PiC �1. According to AIC values, the log link function (AIC: �127) pro- vided a better fit than models with a logit (AIC: �82) or log–log (AIC: �66) link function. In half of the images, the observers were not able to provide a reliable count of the maximum dispersed chromatosomes, necessary to calculate the CI, due to a high level of overlap be- tween the chromatosomes. Above 63 6 9% PiC, individual chro- matosomes overlapped resulting in unreliable CI estimates; above 80 6 9% PiC it was not possible to detect any difference based on CI since all chromatosomes were in the highest cate- gory (CI¼5). No problems were encountered during the estima- tion of PiC, including the darkest images. The observers spent on average 75 6 5 min calculating the CI and only 18 6 9 min de- termining PiC. Results differed significantly among observers for both methods (Friedman’s test: CI: N¼50, df¼2, v2¼11.09, P¼0.04; PiC: N¼50, df¼2, v2¼18.67, P < 0.001), with an average relative standard deviation over all images of 3% for CI and 6% for PiC. Individual regression parameters were similar among the observers (Supplementary Table S1) and the majority of the variation in the PiC estimates was caused by one observer who relied on manual adaptation (N¼23) much more than the other observers (N¼6 and N¼5). Linear regression estimates of PiC values of sRGB versus linearized/normalized images showed that both methods produce concordant results (Supplementary Fig. S2; df¼45, R2¼0.995, P < 0.001, slope¼0.948), indicating that the use of sRGB images did not produce significant system- atic errors in this case. Discussion Animal colouration can be assessed by determining pigment dispersion in individual chromatophores or in multicellular chromatosomes (e.g. [19, 21]). The traditional and widely used CI [20] classifies individual chromatophores or chromatosomes based on their physiological state, indexing their extent of dis- persion. As a result, the CI does not provide information on their morphological state (abundance of pigments). Animals with widely spaced, but fully dispersed, chromatosomes (Fig. 3, D) have, consequently, the same maximum index (CI¼5) as ani- mals with a high abundance and overlap of chromatosomes (Fig. 3, H), even though the difference in darkness is visually ap- parent. This issue has already been considered by Parker [22] who observed catfish with clear differences in darkness, not dis- tinguishable by the values of CI (all falling in the maximum cat- egory). Methods relying on the measurement of the diameter of the chromatosomes [19, 56] have the same problem, since they also omit morphological variation [22]. PiC combines both infor- mation on the distribution and abundance of pigments and is, therefore, able to distinguish physiological differences within the same animal (Fig. 3, A vs. E) and morphological differences between animals with the same physiological chromatosome state (Fig. 3, F vs. G), even in very dark animals (PiC > 80%). The comparison between PiC and CI shows the range where it is possible to transform the values from one method to the other and where PiC is more precise than CI. The logarithmic relation- ship indicates that the more dispersed the pigments are, the more effective the PiC is in detecting small differences between images compared to the CI. Thresholding methods are consid- ered a more reliable tool for image analysis than human judge- ment [44]. Nevertheless, the accuracy and objectivity of PiC is influenced by the amount of manual adaptation applied. The database used during this study consisted of images taken un- der a variety of lighting conditions to show the wide applicabil- ity of PiC. However, automatic thresholding algorithms work best with images taken with identical lighting conditions and camera settings. Manual adaptation of the threshold values, re- quired in cases where the image quality was not optimal (e.g. Fig. 3, B and C), resulted in increased observer variation and subjectivity. In studies where standardization of the images is not possible, extra care should be taken to ensure the objectivity of the study (e.g. observers being made blind to the treatments; between-observer repeatability analysis). These considerations should also be taken into account for the CI. The CI is further- more less precise in darker animals, and it takes up to 4 times longer than PiC. This difference in analysis speed is due to the fact that the CI can only be determined by the manual classifi- cation of every single chromatosome in the image. Moreover, PiC allows testing for transparency, which is important in stud- ies of colour change [21, 57]. Digital photography is a popular technique in animal colour- ation research due to its availability, speed, relative low price and ease of data acquisition [41, 58, 59]. Although there are is- sues with the use of digital images in animal colour studies [41], most of these relate to the control for variation in lightning con- ditions and the conversion of images to animal vision systems [41, 59]. Most cameras produce non-linear images (e.g. sRGB) that generally over- or underestimate light values and rigorous image analysis methods should include linearization and nor- malization of these images [41, 59]. PiC focusses on a priori specified pigments and does not rely on the exact colour or ob- server’s vision system. In this method, the difference between foreground and background pixels in an image is more Figure 4: Relationship between CI and dark PiC fraction. Measurements were performed on 50 images of C. crangon (Fig. 3). Mean values and SD for the read- ings of three observers are given per image. The solid line shows the beta regres- sion fit (with log link function). 6 | Siegenthaler et al. Deleted Text: pigment cover Deleted Text: chromatophore index Deleted Text: . Deleted Text: . Deleted Text: . Deleted Text: P Deleted Text: Deleted Text: - Deleted Text: - Deleted Text: - Deleted Text: - Deleted Text: pigment cover Deleted Text: T Deleted Text: . Deleted Text: . Deleted Text: X Deleted Text: . Deleted Text: . Deleted Text: X2 Deleted Text: s Deleted Text: . Deleted Text: . Deleted Text: DISCUSSION Deleted Text: [ Deleted Text: Chromatophore Index Deleted Text: , Deleted Text: , Deleted Text: , Deleted Text: , Deleted Text: , Deleted Text: & Deleted Text: s Deleted Text: , Deleted Text: , Deleted Text: which Deleted Text: - Deleted Text: s Deleted Text: PiC focus Deleted Text: ses important than the exact colour, thus stable lighting conditions are less relevant for PiC than for methods requiring linearized images. Studies that analyse chromatophores and pigment mi- gration (see Table 1 for examples) usually focus on a limited number of pigments, in high contrast with the background. In these types of studies, PiC can be used also with sRGB images (as shown by the concordant PiC values of sRGB and linearized images reported above) as long as the users are aware of the limitations of the use of non-linear images. In cases where a more precise, objective and rigours determination of animal col- our is required, image normalization and standardization can be performed prior to PiC determination. Standardization of lighting conditions and camera settings is also advised in these cases. Besides being less constrained regarding lighting condi- tions, PiC is also easy to use and fast in the analysis of large sur- faces (opposed to spectrometry; [3]). The study of animal colouration is a broad field of investiga- tion encompassing molecular, cellular, physiological, behaviou- ral and evolutionary questions [1, 12]. The proposed methodology combines the advantages of digital image acquisi- tion with the power of a free open-source program. PiC is simple to use, can also be easily employed for educational purposes [60] and can be applied in any system where rapid colour change is determined by pigment migration in chromatophores. The brown shrimp’s chromatosomes system is a widely appli- cable model since its physiological factors are well studied and its pigment system is complex and essentially similar to those of vertebrates [8, 13, 61–63]. The proposed method will thus be a useful tool in future investigations on animal colouration as a fast and effective proxy for the interpretation of complex and dynamic biological systems in a wide range of species. Acknowledgements We would like to thank Asma Althomali, Clément Dufaut and Héctor Abarca Velencoso for their great help with image analysis; Paul Oldfield and Steve Manning for their assis- tance with the sampling and Alexander Mastin for his ad- vice on statistical analyses. Three anonymous reviewers have provided insightful and valuable comments and feed- back to a previous version of this manuscript. This project has been funded by the University of Salford and the Mersey Gateway Environmental Trust. Conflict of interest statement. None declared. Supplementary data Supplementary data is available at Biology Methods and Protocols online. References 1. Stevens M, Rong CP, Todd PA. Colour change and camouflage in the horned ghost crab Ocypode ceratophthalmus. Biol J Linn Soc 2013;109:257–70. 2. Stevens M, Stoddard M, Higham J. Studying primate color: to- wards visual system-dependent methods. Int J Primatol 2009;30:893–917. 3. White TE, Dalrymple RL, Noble DWA et al. Reproducible re- search in the study of biological coloration. Anim Behav 2015;106:51–7. 4. Holmes W. The colour changes and colour patterns of Sepia officinalis L. Proc Zool Soc Lond 1940;A110:17–35. 5. Hanlon RT. Cephalopod dynamic camouflage. Curr Biol 2007;17:400–4. 6. Taylor CH, Gilbert F, Reader T. Distance transform: a tool for the study of animal colour patterns. Methods Ecol Evol 2013;4:771–81. 7. Merilaita S, Dimitrova M. Accuracy of background matching and prey detection: predation by blue tits indicates intense selection for highly matching prey colour pattern. Funct Ecol 2014;28:1208–15. 8. Koller G. €Uber chromatophorensystem, farbensinn und farb- wechsel bei Crangon vulgaris. J Comp Physiol A 1927;5:191–246. 9. Perkins EB, Snook T. Control of pigment migration in the chromatophores of crustaceans. Proc Natl Acad Sci USA 1931;17:282–85. 10. Darnell MZ. Ecological physiology of the circadian pigmenta- tion rhythm in the fiddler crab Uca panacea. J Exp Mar Biol Ecol 2012;426–27:39–47. 11. Sköld HN, Aspengren S, Wallin M. Rapid color change in fish and amphibians - function, regulation, and emerging applica- tions. Pigment Cell Melanoma Res 2013;26:29–38. 12. Umbers KDL, Fabricant SA, Gawryszewski FM et al. Reversible colour change in Arthropoda. Biol Rev 2014;89:820–48. 13. Elofsson R, Kauri T. The ultrastructure of the chromatophores of Crangon and Pandalus (Crustacea). J Ultrastruct Res 1971;36:263–70. 14. Bauer RT. Color patterns of the shrimps Heptacarpus pictus and H. paludicola (Caridea: Hippolytidae). Mar Biol 1981;64:141–52. 15. Fujii R. The regulation of motile activity in fish chromato- phores. Pigment Cell Res 2000;13:300–19. 16. McNamara JC. Morphological organization of crustacean pig- mentary effectors. Bio Bull 1981;161:270–80. 17. Meelkop E, Temmerman L, Schoofs L et al. Signalling through pigment dispersing hormone-like peptides in invertebrates. Prog Neurobiol 2011;93:125–47. 18. Smith HG. The receptive mechanism of the background re- sponse in chromatic behaviour of crustacea. Proc R Soc B 1938;125:250–63. 19. Peter J, Meitei KV, Ali AS et al. Role of histamine receptors in the pigmentary responses of the wall lizard, Hemidactylus fla- viviridis. Curr Sci 2011;101:226–29. 20. Hogben L, Slome D. The pigmentary effector system. VI. The dual character of endocrine co-ordination in amphibian col- our change. Proc R Soc B 1931;108:10–53. 21. Auerswald L, Freier U, Lopata A et al. Physiological and mor- phological colour change in Antarctic krill, Euphausia superba: a field study in the Lazarev Sea. J Exp Biol 2008;211:3850–58. 22. Parker GH. Methods of estimating the effects of melanophore changes on animal coloration. Bio Bull 1943;84:273–84. 23. Flores EE, Chien Y-H. Chromatosomes in three phenotypes of Neocaridina denticulata Kemp, 1918: morphological and chro- matic differences measured non-invasively. J Crust Biol 2011;31:590–97. 24. Ali SA, Naaz I. Understanding the ultrastructural aspects of berberine-induced skin-darkening activity in the toad, Bufo melanostictus, melanophores. J Microsc Ultrastruct 2015;3:210–19. 25. Ali SA, Salim S, Sahni T et al. 5-HT receptors as novel targets for optimizing pigmentary responses in dorsal skin melano- phores of frog, Hoplobatrachus tigerinus. Br J Pharmacol 2012;165:1515–25. 26. Gao L, Yu Z, Meng D et al. Analogue of melanotan II (MTII): a novel melanotropin with superpotent action on frog skin. Protein Pept Lett 2015;22:762–66. Animal pigment cover quantification | 7 Deleted Text: t Deleted Text: s Deleted Text: [ Deleted Text: also Deleted Text: see: 27. Belden LK, Blaustein AR. UV-B induced skin darkening in lar- val salamanders does not prevent sublethal effects of expo- sure on growth. Copeia 2002;2002:748–54. 28. Eagleson GW, Selten MM, Roubos EW et al. Pituitary melano- trope cells of Xenopus laevis are of neural ridge origin and do not require induction by the infundibulum. Gen Comp Endocrinol 2012;178:116–22. 29. Eagleson GW, van der Heijden RA, Roubos EW et al. A develop- mental analysis of periodic albinism in the amphibian Xenopus laevis. Gen Comp Endocrinol 2010;168:302–06. 30. Gouveia GR, Lopes TM, Neves CA et al. Ultraviolet radiation induces dose-dependent pigment dispersion in crustacean chromatophores. Pigment Cell Res 2004;17:545–48. 31. Wilcockson DC, Zhang L, Hastings MH et al. A novel form of pigment-dispersing hormone in the central nervous system of the intertidal marine isopod, Eurydice pulchra (leach). J Comp Neurol 2011;519:562–75. 32. Meelkop E, Marco HG, Janssen T et al. A structural and func- tional comparison of nematode and crustacean PDH-like se- quences. Peptides 2012;34:74–81. 33. Marco HG, G€ade G. Biological activity of the predicted red pigment-concentrating hormone of Daphnia pulex in a crustacean and an insect. Gen Comp Endocrinol 2010;166:104–10. 34. Ali SA, Meitei KV. Nigella sativa seed extract and its bioactive compound thymoquinone: the new melanogens causing hyperpigmentation in the wall lizard melanophores. J Pharm Pharmacol 2011;63:741–46. 35. Jiang Q, Wong AO. Signal transduction mechanisms for auto- crine/paracrine regulation of somatolactin-alpha secretion and synthesis in carp pituitary cells by somatolactin-alpha and -beta. Am J Physiol Endocrinol Metab 2013;304:E176–86. 36. Xu J, Xie FK. Alpha- and beta-adrenoceptors of zebrafish in melanosome movement: a comparative study between em- bryo and adult melanophores. Biochem Biophys Res Commun 2011;405:250–55. 37. Lennquist A, Martensson Lindblad LG, Hedberg D et al. Colour and melanophore function in rainbow trout after long term exposure to the new antifoulant medetomidine. Chemosphere 2010;80:1050–55. 38. Yoshikawa N, Matsuda T, Takahashi A et al. Developmental changes in melanophores and their asymmetrical respon- siveness to melanin-concentrating hormone during meta- morphosis in barfin flounder (Verasper moseri). Gen Comp Endocrinol 2013;194:118–23. 39. Brown FA, Sandeen MI. Responses of the chromatophores of the fiddler crab, Uca, to light and temperature. Physiol Zool 1948;21:361–71. 40. Nguyen N, Sugimoto M, Zhu Y. Production and purification of recombinant somatolactin beta and its effects on melano- some aggregation in zebrafish. Gen Comp Endocrinol 2006;145:182–87. 41. Stevens M, P�arraga CA, Cuthill IC et al. Using digital photogra- phy to study animal coloration. Biol J Linn Soc 2007;90:211–37. 42. Tajima R, Kato Y. Comparison of threshold algorithms for au- tomatic image processing of rice roots using freeware ImageJ. Field Crops Res 2011;121:460–63. 43. Swanson J-D, Carlson JE, Guiltinan MJ. Use of image analysis software as a tool to visualize non-radioactive signals in plant in situ analysis. Plant Mol Biol Rep 2006;24:105. 44. Drury JA, Nik H, van Oppenraaij RH et al. Endometrial cell counts in recurrent miscarriage: a comparison of counting methods. Histopathology 2011;59:1156–62. 45. Schneider CA, Rasband WS, Eliceiri KW. NIH Image to ImageJ: 25 years of image analysis. Nat Methods 2012;9:671–75. 46. Hartig SM. Basic image analysis and manipulation in ImageJ. Curr Protoc Mol Biol 2013. doi: 10.1002/0471142727.mb1415s102. 47. Ferreira T, Rasband W. ImageJ User Guide. ImageJ/Fiji 1.46. Bethesda, Maryland: National Institutes of Health, 2012. 48. Brocher J. Qualitative and guantitative evaluation of two new histogram limiting binarization algorithms. IJIP 2014;8:17–65. 49. Jensen EC. Quantitative analysis of histological staining and fluorescence using ImageJ. Anat Rec 2013;296:378–81. 50. Papadopulos F, Spinelli M, Valente S et al. Common tasks in microscopic and ultrastructural image analysis using ImageJ. Ultrastruct Pathol 2007;31:401–07. 51. Brown FA, Wulff VJ. Chromatophore types in Crago and their endocrine control. J Cell Comp Physiol 1941;18:339–53. 52. Ridler TW, Calvard S. Picture thresholding using an iterative selection method. Ieee Trans Syst Man Cybern 1978;8:630–32. 53. Landini G. Auto Threshold (ImageJ). http://fiji.sc/Auto_ Threshold (7 March 2016, date last accessed). 54. Kapur JN, Sahoo PK, Wong AKC. A new method for gray-level picture thresholding using the entropy of the histogram. Comput Vision Graph 1985;29:273–85. 55. Cribari-Neto F, Zeileis A. Beta regression in R. J Stat Soft 2010;34:1–24. 56. McNamara JC, Ribeiro MR. The calcium dependence of pig- ment translocation in freshwater shrimp red ovarian chro- matophores. Biol Bull 2000;198:357–66. 57. Sköld HN, Svensson PA, Zejlon C. The capacity for internal colour change is related to body transparency in fishes. Pigment Cell Melanoma Res 2010;23:292–95. 58. Stevens M, Merilaita S. Animal camouflage: current issues and new perspectives. Philos Trans R Soc B 2009;364:423–27. 59. Troscianko J, Stevens M. Image calibration and analysis tool- box - a free software suite for objectively measuring reflec- tance, colour and pattern. Methods Ecol Evol 2015;6:1320–31. 60. Heggland SJ, Lawless AR, Spencer LW. Visualizing endocrinol- ogy in the classroom. Am Biol Teach 2000;62:597–601. 61. Keeble F, Gamble FW. The colour-physiology of higher crusta- cea. Philos Trans R Soc B 1904;196:295–388. 62. Elofsson R, Hallberg E. Correlation of ultrastructure and chemical composition of crustacean chromatophore pig- ment. J Ultrastruct Res 1973;44:421–29. 63. Rao KR. Crustacean pigmentary-effector hormones: chemis- try and functions of RPCH, PDH, and related peptides. Am Zool 2001;41:364–79. 8 | Siegenthaler et al. http://fiji.sc/Auto_Threshold http://fiji.sc/Auto_Threshold bpx003-TF1