Ethics statement
All protocols and animal care procedures were performed according to local and international ethical guidelines (NIH Publications No. 8023, revised 1978) and were approved by the Brazilian environment agency (SISBIO 32763-3) and by the Animal Research Ethics Committee (CEPAE) of the National Primate Center, in Pará, Brazil (CENP-PA) (CEPAN n°33/2011). The monkeys were not harmed during the study.
Subjects
Six Alouatta spp. individuals, housed at the Brazilian National Primate Center (CENP-PA) were trained for the behavioural tests: four adult males (three A. caraya and one A. belzebul) and two females, one adult (A. seniculus) and one juvenile (A. belzebul) (Fig. 3). Subjects had no previous history of reinforcement and some were resistant, as they were rescued from wildlife, from distress conditions, habitat loss or hunter’s captivity, and were not used to human handling. Animals were kept in enclosures (2.5 m wide × 4 m deep × 2.3 m high) connected through a window to twin enclosures. The enclosures were exposed to normal fluctuations in photoperiod, with a cycle of approximately 12 h daylight and 12 h dark. Subjects were fed twice a day with primate ration and fruits, leaves daily, juice or milk twice a week, and water was provided ad libitum. The subjects were not food deprived prior to the experiment, having it removed only during the test sessions.
Blood samples collection
Blood samples were collected by specialized staff from the CENP-PA. The procedure followed local routine with anesthetic association of tiletamine and zolazepam (10 mg/kg). Approximately 6 ml of blood was collected from the femoral vein with a 22G hypodermic needle adapted to a BD Vacutainer® tube with 3.2% buffered sodium citrate solution. After blood collection, animals were returned to an empty enclosure until full recovery.
Genetic analysis
DNA extraction, PCR, and sequencing
Genetic analyzes were performed to investigate the lws, mws, and sws1 opsin genes of two Alouatta species, A. seniculus and A. caraya. Genomic DNA was extracted from the blood samples using standard protocols (Puregene DNA®, Gentra System). Polymerase chain reactions (PCRs) were carried out to amplify exons 3 and 5 of the lws and mws opsin genes, and the exon 1 of the sws1 opsin gene, to identify the specific amino acid residues responsible for the opsins spectral tuning. The primers used to amplify the lws/mws and the sws1 exons were described elsewhere [18, 37]; (Supplementary Table S1). PCRs were performed using High Fidelity Platinum Taq Polymerase, and PCR conditions are described in the Supplementary Table S2. The PCR products were visualized by agarose gel electrophoresis (2%), and purified using the Illustra™ GFX PCR DNA and Gel Band purification kit (GE Healthcare), according to the manufacturer protocol. The purified samples were directly sequenced in both directions, using BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystem™) and a 3500xL sequencer (Applied Biosystem™), and sequences were analyzed with BioEdit [38].
Molecular cloning and inter-exon PCR
Based on the variations observed from the sequencing results, we confirmed that both opsin genes mws and lws were amplified together by the primer pairs used. In order to sequence the two genes separately, PCR products obtained from amplification of exon 5 from the two samples (A. seniculus and A. caraya), were cloned into plasmid vectors using TA cloning® kits (Invitrogen). Multiple clones of each sample were isolated, purified via spin columns (Qiagen), and sequenced using M13 vector primers. Based on sequences analyses from 10 individual clones of each sample, we were able to differentiate exon 5 of the mws and the lws genes, and to design specific primers for each gene. In an inter-exon PCR, we combined a common forward primer for both genes located in exon 3, with reverse primers specific for the mws and for the lws genes, located at exon 5 (Supplementary Table S1). The PCRs were performed using the enzyme TaKaRa LA Taq® (Takara Bio USA), and PCR parameters were based on Davidoff, Neitz & Neitz [39] (Supplementary Table S2). The PCR products over 2 kb were visualized using a 2% agarose gel, purified, and sequenced in both directions as described above. Sequences were analyzed with BioEdit 7.25 [38], and exons 3 and 5 of each gene were identified.
Estimates of the absorption peaks of the LWS, MWS, and SWS1 opsins
The opsin sequences were aligned with the corresponding sequences of other primates and spectral tuning sites were identified to estimate the opsins λmax. For the SWS1 opsin, amino acids were numbered based on the bovine rhodopsin (RH1) (GenBank accession number NM_001014890), and residues 46, 49, 52 81, 86, 90, 93, 114, 116, and 118, were considered for spectral tuning estimates [40]. For the LWS and MWS opsins, sequences were aligned with human LWS opsin to identify the spectral tuning sites 180, 277, and 285, and estimate the opsins λmax [22, 41,42,43].
Behavioural tests
Apparatus
The behavioural colour discrimination evaluation was performed in a computerized system consisting of a video monitor (Diamond Pro 2070SB, Mitsubishi, Cypress, CA, USA, spatial resolution: 2048 × 1536 at 86 Hz -, frame rate: 160 Hz, colour resolution: 14 bits) in which the stimuli were displayed on a video monitor adapted to receive touch inputs. A software programmed the experimental sessions and recorded the monkey’s responses. The setup (ViSaGe system® - Cambridge Research Systems, Rochester, UK, and CRT monitor), previously used and described in detail by Goulart and collaborators [20], was adapted for Alouatta subjects, with a larger food dispenser and tray. Gamma correction was performed to calibrate the luminance of the monitor guns using the software VSG Desktop (CRS) and a ColorCAL MKII Colorimeter (Cambridge Research Systems, Rochester, UK). The monitor was placed at a distance of 20 cm from the subjects’ eyes, allowing them to reach the target only with the fingers. For this experiment, the equipment was mounted on a mobile rack, allowing in situ experiments, attached to a 60 × 60 × 60 cm experimental chamber.
Stimulus
The stimulus was an adaptation by Goulart and collaborators [24], of the Cambridge Colour Test (CCT) (Cambridge Research Systems, Rochester, UK), in which the Landolt-C target is replaced by a composite image resembling a 5 cm square patch that can appear in one of four different positions (top left, top right, bottom left, bottom right). Both, target and background, were composed of a matrix of circles of random diameter, ranging from 5 to 9 mm, and random luminance, ranging from 7 and 15 cd/m2. The CIE 1976 chromaticity diagram coordinates for the background chromaticity were u’ = 0.1977, v’ = 0.4689 for the first ellipse (E1), u’ = 0.1925, v’ = 0.5092, for the second ellipse (E2) and u’ = 0.2044, v’ = 0.4160 for the third ellipse (E3). In each trial, target chromaticity changed dynamically, with the target varying within one of the 20 predefined chromatic vectors. We used a staircase procedure to estimate colour discrimination threshold for each vector. The excursion along each vector varied between 1100 × 10− 4 and 20 × 10− 4 u’v’ units, getting one step closer to the background chromaticity for every correct response and one step away for wrong responses. Discrimination thresholds, expressed as threshold vector length in u’v’ units in 1976 CIE chromaticity diagram, were obtained along 20 equally spaced chromatic vectors surrounding the reference chromaticity, by an average of the last seven of 11 reversals. We considered reversals wrong responses followed by correct ones, and correct responses followed by wrong ones. MATLAB routines were used to interpolate these thresholds resulting in an ellipse around the background chromaticity.
Procedures
General procedures
All subjects were trained 3 days a week via positive reinforcement through a shaping procedure (Fig. 4). Session duration was up to an hour. They were first adapted to the presence of the researcher, to the type of reinforcement – banana flavored pellets (Bioserv®, Flemington USA), grapes or nuts – and to the test apparatus. Following this phase, subjects were trained through several steps with increasing complexity from touching the monitor to a stage where they consistently discriminated a colour patch from the background. Each trial began with the presentation of the stimulus and ended with its disappearance, after a touch response hit any point on the screen. Touch responses on the target area were immediately followed by reward presentation. A 5 s intertrial interval (ITI) was programmed before the following stimulus presentation. Touch responses on any other part of the screen were followed by the ITI alone.
Training phase
During training, target chromaticities were highly saturated distant from any confusion lines predicted for dichromats in the CIE 1976 chromaticity diagram, deliberately facilitating target detection for training purposes. At the beginning of the training, the target-background distance was kept constant at the maximum (1100 × 10− 4 u’v’ units). When subjects learned to consistently touch over the target, we introduced the dynamic staircase variation of the target-background distance, with the minimum distance of target to background chromaticity defined as 200 × 10− 4 u’v’ units. The minimum chromatic target-background distance gradually changed after each successful training day from the easy training parameters (minimum distance of 200 × 10− 4 u’v’ units) to the test parameters (minimum distance of 20 × 10− 4 u’v’ units). Once subjects reached three consecutive days of consistent discrimination at the minimum distance, the training was considered successful, and they received a pretest session that simulated the test conditions. In the test simulation, subject was presented with three novel vectors (outside hypothesized confusion lines), and the minimum distance was set at 20 × 10− 4 u’v’ units, which would eventually lead to failure to distinguish target from background, at the threshold region.
Testing phase
Subjects were tested along 20 chromatic vectors for each ellipse background chromaticity, defined around the same background coordinates used by the Cambridge Colour Test. A test session presented five out of 20 vectors, and the whole test consisted of four sessions for each background chromaticity. The staircase was automatically generated, with upper and lower limits respectively at 1100 × 10− 4 u’v’ units and 20 × 10− 4 u’v’ units. When the subjects’ performance reached the criterion of 11 reversal trials, a discrimination threshold for that vector was derived from the average distances between the background chromaticity and the target chromaticities at which the last seven reversals occurred. Chromatic discrimination thresholds coordinates were fitted to an ellipse model and the parameters of the ellipses (major axis length, axis ratio, and angle) guided the behaviourally based phenotypic classification.