Overview of experimental design
To test for and characterise among-individual variation in N. heteropoda we aimed to subject each of the 48 individuals to three repeats of two separate behavioural assays: an Open field trial (OFT) and a Food and shelter trial (FST). The order of assays was held constant, with all individuals completing 3 x OFT followed by 3 x FST over a five-week period, with a minimum of 48 hours between any two successive trials. This design would have yielded 288 trials (48 individuals x3 repeats x2 assays). However, some mortalities occurred during the investigation period. In some cases, we opportunistically replaced mortalities with new stock shrimp such that our final data set analysed actually comprised 273 trials on 53 shrimp, with a mean of 2.5 OFT and 2.6 FST per individual.
Trials were run between 0900 and 1300 hours with individuals tested in a random order. On any given day, all trials conducted in the laboratory were of a single assay type, with duplicate experimental arena tanks allowing two shrimp to be tested simultaneously. These experimental tanks (30 cm x 20.5 cm x 21cm) were filled to 5cm3with water from the sump of the recirculating water system to which all individual housing units were connected. The water was changed every 6 trials to reduce any influence of conspecific cues that might be produced. Each tank was filmed from above using a Sunkwang C160 video camera mounted with a 5-50mm manual focus lens, and the tracking software Viewer II (BiObserve) used to measure behaviours putatively linked to shy-bold variation (described below for each assay). The experimental tanks were surrounded by cardboard screens to exclude external visual stimuli that might otherwise impact behaviour.
At the end of each trial, size and sex data were recorded and the shrimp was then returned to individual housing. For size we measured both length (from the tip of the rostrum to the posterior end of the telson using digital callipers to the nearest 0.01mm) and mass (using a digital balance after dabbing the animal with a tissue to remove water droplets to the nearest 0.001g). However, length and mass measures were highly correlated across observations (r=0.717, t 268 = 16.843, P<0.001), while mass was slightly more repeatable at the individual level (Rmass= 0.768 versus, Rlength= 0.581). Assuming size was (approximately) constant for individuals over study period then this suggests mass has a lower measurement error. We therefore used mass as our measure of size in all analyses. Sex was estimated from external morphology after each trial, and shrimp were scored as a male, female, or of unknown sex. On average, females are larger and more opaque than males, and have more rounded bellies and body plates . Females can sometimes be seen carrying eggs in their swimmerets and may show a distinctive ‘saddle’ marking . Given uncertainty in sex determination, we elected to score it after each trial blind to any previous assessments of the same individual. After all observations were complete, we assign a single sex determination of a male or female if ≥5/6 trial-specific assessments were in agreement. We assigned sex as ‘unknown’ if this criterion was not met.