Nanoscale ultrastructures increase the visual conspicuousness of signalling traits in obligate cleaner shrimps
Data files
Aug 20, 2024 version files 18.58 KB
-
CombinedSphereMeasurements.csv
-
README.md
-
Reflectance_calibrated_photography.csv
-
SimulationData.csv
Abstract
To be effective, signals must be detectable; thus, signal theory predicts that organisms should evolve signals that are conspicuous to intended receivers in natural signalling environments. Cleaner shrimps remove ectoparasites from reef fish clients and many signal their intent to clean by whipping their long, white antennae. Since white is a reliably conspicuous colour in aquatic environments, we hypothesized that selection has acted to increase broad-spectrum antennal reflectance in these species. We compared reflectance and antennae ultrastructure in three obligate cleaner species with one facultative cleaner and one non-cleaner in two families (Palaemonidae and Lysmatidae). Obligate cleaner antennae had a reflectance of 41-52% of human-visible light, compared to 35% in the facultative cleaner and 21% in the non-cleaner. Scanning electron microscopy (SEM) revealed the antennae in two obligate cleaner species contain a layer of densely packed, high refractive index spheres (300-400nm diameter), which optical models showed are effective at increasing reflectance. In the third obligate cleaner, SEM showed thick (~6µm) chitinous layers that may also increase reflectance. The facultative and non-cleaning species had no visible antennae ultrastructure beyond chitinous exoskeleton. Our results suggest that some obligate cleaners have evolved ultrastructural modifications that increase the conspicuousness of the antennae as signals.
README: READ ME GUIDE TO SUPPLEMENTAL DATA AND CODE FILES
Caves et al., Nanoscale ultrastructures increase the visual conspicuousness of signalling traits in obligate cleaner shrimps. Journal of Experimental Biology.
R_code.R:
-An R script which runs all of the analyses and creates all of the figures in the manuscript and extended data.
-Three files are necessary to run the complete R script:
1) Reflectance_calibrated_photography.csv
: this file contains reflectance data generated using calibrated color photography, for various species of shrimps
Columns are:
-Species: Shrimp species
-Specimen: a number indicating which individual of a given species reflectance values are taken from
-Measurement: multiple measurements were taken per individual; this column indicates which measurement number a reflectance value is associated with
-Reflectance: measured reflectance in percent
-Type: Shrimp type (either “cleaner”, “facultative cleaner”, “or non cleaner”)
-Category: A categorical variable for the purposes of statistically comparing reflectance in cleaners versus others (either “cleaner” or “non cleaner”)
2) CombinedSphereMeasurements.csv
: this file contains measurements (taken from Scanning Electron Microscope images) of the width of spheres in Ancylomenes pedersoni and Lysmata amboinensis
Columns are:
-Specimen: A unique identifier for each specimen measured
-Width: the width of the measured sphere; multiply by 1000 for nanometers
-Species: either AP (Ancylomenes pedersoni) or LA (Lysmata amboinensis)
3) SimulationData.csv
: this file contains data output from FDTD simulations run in Lumerical, specifically simulated reflectance varying refractive index, layer thickness, and sphere diamter (details in main text)
Columns are:
-RI_Diam: Sphere diameter in simulations varying diameter (see Figure 2D and main text)
-RI_1_57_adj: simulated reflectance with varying sphere diameter, with refractive index of 1.57 (chitin); see Figure 2D and main text
-RI_1_78_adj: simulated reflectance with varying sphere diameter, with refractive index of 1.78 (isoxanthopterin); see Figure 2D and main text
-RI_2_00_adj: simulated reflectance with varying sphere diameter, with refractive index of 2.00 (pteridines); see Figure 2D and main text
-Thickness: Layer thickness in simulations varying layer thickness (see Figure 2C and main text)
-Thickness_1_57_adj: simulated reflectance with varying layer thickness, with refractive index of 1.57 (chitin); see Figure 2C and main text
-Thickness_1_78_adj: simulated reflectance with varying layer thickness, with refractive index of 1.78 (Isoxanthopterin); see Figure 2C and main text
-Thickness_2_00_adj: simulated reflectance with varying layer thickness, with refractive index of 2.00 (pteridines); see Figure 2C and main text
-Presence_RI: refractive index for simulations varying refractive index (see Figure 2B and main text)
-Presence_Refl_adj: simulated reflectance with varying refractive index (see Figure 2B and main text)
Methods
Reflectance Measurements
We used calibrated photography under a stereomicroscope to measure reflectance from the antennae following methods in [22]. Calibrated photography was used instead of spectoradiometry because cleaner shrimp antennae are cylindrical rather than flat, causing light to scatter in directions that are not captured by the fibre optic sensor, likely underestimating reflectance. In brief, photographs of antennae were taken with an iPhone (model 13, Apple Inc., Cupertino, CA) and included a reflectance standard comprising five grey paint swatches of measured reflectance. We measured the 0-255 value from each swatch of the grey standard (using the image analysis software Fiji [23]), using only the green channel of the sRGB image. The green channel was used both to approximate as closely as we are able brightness vision in particular, and because of the three colour channels, it is most closely aligned with the ambient illuminant on a coral reef [24]. We then plotted the green values of the grey standards against their known reflectance values, and fit exponential equations to these values to create a calibration equation for each photograph. Then, in Fiji, we measured green-channel values from four to ten regions on each antenna; the number of regions sampled did not affect our results, as randomly sampling only 4 regions from each specimen and re-running our analyses did not change our conclusions in any instance. We then converted these antennae values to reflectance using the calibration equations generated from the grey standards. We averaged reflectance values across individuals to yield a single measure of antennal reflectance for each specimen, and averaged across specimens to yield average values for each species, for use in analysis.
Scanning Electron Microscopy
To identify underlying structural features that may enhance antennae reflectance we used SEM. Two 3mm long sections of antenna per animal were excised from two specimens per species and fixed for 12 hours in 2.5% glutaraldehyde buffered with artificial seawater. After fixation, each sample went through a dehydration series of 30%, 30%, 50%, 50%, 70%, 70%, 90%, 90%, 100%, 100%, 100% EtOH (15 minutes/step). Samples were then dried using a LADD CPD3 critical point dryer (Ladd Research Industries, Williston, VT, USA) to preserve tissue ultrastructure. Once dried, samples were freeze-fractured to expose the cross-section, mounted on aluminium SEM stubs with copper tape, and sputter-coated with an ~8nm thick layer of gold (Denton Desk V; Denton Vacuum LLC, Moorestown, NJ, USA). The samples were imaged using an Apreo S scanning electron microscope (ThermoFisher Scientific, Waltham, MA, USA) at the Duke University Shared Materials and Instrumentation Facility with an acceleration voltage of 1kV and magnifications of 2500x–15000x. We used Fiji [25] for morphometric analyses of the sphere layer found using SEM.
Optical Modelling
In two of the three obligate cleaner species, SEM images showed a layer of spherical nanoparticles inside the cuticle layer. To investigate how the morphology and optical properties of these particles impact antennae reflectance, we performed finite-difference time-domain simulations (FDTD) using the Lumerical solver version 2020b (Ansys, Canonsburg, PA, USA). Random close-packed aggregations of spheres mimicking the arrangement observed in the antennae were generated using the Uniform Random Particle Distribution (URPD) structure in Lumerical.
We performed three sets of simulations to determine the effects of sphere refractive index, sphere layer thickness, and sphere layer diameter on reflectance. First, to determine the effect of sphere refractive index on reflectance, we simulated a 2µm thick layer of non-absorbing spheres underneath a 5µm layer of homogenous chitin, varying sphere refractive index from 1.57 (approximately that of chitin) to 2.00 (a lower estimate of that of the pteridine granules found in white Pierid butterflies [11]) in increments of 0.01. Notably, this range encompassed the refractive index of isoxanthopterin (1.78), which prior work has identified as the molecular basis of the spheres underlying white body colouration in Lysmata amboinensis (one of the obligate cleaners studied here) [12]. Second, we simulated layers of spheres (randomly assigned diameters between 300-400nm, based on what was found in A. pedersoni and L. amboinensis) between 1µm and 10µm thick in 300nm thickness intervals to determine how reflectance changes with layer thickness. Third, we simulated the effects of sphere diameter on the reflectance of a 2µm thick layer of spheres (again based on what we observed in A. pedersoni and L. amboinensis) ranging between 100nm and 1000nm in diameter, in 30nm intervals; within each simulation, all spheres had the same diameter (to within 5nm of one another). For investigating the effects of both layer thickness and sphere diameter, we simulated three different sphere refractive indices, 1.57 (chitin), 1.78 (isoxanthopterin), and 2.00 (pteridine), with a background matrix equal to the refractive index of seawater (1.34).
All simulations were performed in a 4µm x 4µm x 12µm domain with periodic boundary conditions in the x and y directions, broadband (400nm – 700nm) plane wave source propagating in the z direction, and perfectly-matched layer boundary conditions in the z direction (a computational representation of open boundaries that does not reflect any light at the edges of the domain). Although we simulated one sphere layer, antennae are effectively one layer wrapped into a cylinder, meaning that most incident light upon the antennae passes through two layers. Therefore, we report reflectance as: R + RT2, where R is the reflectance of one layer and T is the transmittance, as this equation accounts for reflectance of the front layer, plus reflectance of the back layer after having passed through the front layer twice.