Social experience early in life appears to be necessary for the development of species-typical behavior. Although isolation during critical periods of maturation has been shown to impact behavior through gene expression and brain development in invertebrates and vertebrates, workers of some ant species appear resilient to social deprivation and other neurobiological challenges that occur during senescence or due to loss of sensory input. It is unclear if and to what degree neuroanatomy, neurochemistry, and behavior will show deficiencies if social experience in the early adult life of worker ants is compromised. We reared workers of Camponotus floridanus from adult eclosion under conditions of social isolation for two to 53 days, quantified brain compartment volumes, recorded biogenic amine levels in individual brains, and evaluated movement and behavioral performance to compare the neuroanatomy, neurochemistry, brood-care behavior, and foraging (predatory behavior) of isolated workers with that of workers experiencing natural social contact after adult eclosion.  We found that the volume of the antennal lobe, which processes olfactory inputs, was significantly reduced in workers isolated for an average 40 days, whereas the size of the mushroom bodies, centers of higher-order sensory processing, increased after eclosion and was not significantly different from controls. Titers of the neuromodulators serotonin, dopamine, and octopamine remained stable and were not significantly different in isolation treatments and controls. Brood care, predation, and overall movement were reduced in workers lacking social contact early in life. These results suggest that the behavioral development of isolated workers of C. floridanus is specifically impacted by a reduction in the size of the antennal lobe. Task performance and locomotor ability therefore appear to be sensitive to a loss of social contact through a reduction of olfactory processing ability rather than change in the size of the mushroom bodies, which serve important functions in learning and memory, or the central complex, which controls movement.

Colony collection and culture

Three queenright colonies of C. floridanus were collected in or near Gainesville, Florida in 2012 and reared in Harris environmental chambers at 25°C and 40-55% relative humidity on a 12:12 light cycle.  Parent colonies from which workers were sampled had a minimum size of 100, equally distributed between minors and majors.  Artificial nests were made from test tubes 75% filled with water and fitted with a tight cotton plug and placed in Fluon®-lined (Bioquip, Rancho Dominguez, CA) plastic nest boxes 16x11x6 cm to 32x22x6 cm depending on colony size.  Colonies were fed 1M sucrose on saturated cotton balls and varied protein sources (mealworms, scrambled eggs, and fruit flies) every other day in rotation.

Worker social isolation

C. floridanus has a dimorphic worker caste.  We used minor workers because of their diverse behavioral repertoire, which includes brood care and foraging, rather than majors, which are limited to defense.  Four groups of callow minors were collected from each of three parent colonies.  Randomly selected callows were isolated from nestmates for 2-3 days (n=29), 5-6 days (n=36), 20-22 days (n=44), and 40 days (n=29, mean = 40.08 days, range = 30 – 53 days; herein 40 days) in separate Fluon®-lined boxes (16 x 11 x 6 cm).  Socially isolated workers (“isolates”) were provided similar conditions in smaller nest boxes and the same diet as parent colonies.  After the prescribed isolation period, isolates were assayed for behavioral performance and subsequently sacrificed for immunohistochemical analysis of brain structure, or for quantification of brain biogenic amine titers.  Callow (n = 29) and mature minor workers (“matures”, n = 46) were collected and assayed using the same methods.  Callow and mature controls served as comparative groups for 2-day isolates and 40-day isolates, respectively, and we report primarily on differences in behavior, neuroanatomy, and neurochemistry between these groups.  We compared brain volumes of callow controls estimated to be 2 days old and 2--3-day isolates to ensure there were no intrinsic differences in brain size post eclosion.  We used 5- and 20-day isolates lacking age-matched controls to estimate the trajectory of behavioral and neural development from 2 days to 53 days of isolation (Supp Fig 1-4).  Isolated workers had a survival rate of 63.7%.  Because of the relatively high mortality of workers experiencing longer periods of isolation, sample sizes for 40-day isolates are generally lower.  Head width, measured as the widest distance across the eyes, served as a proxy for body size.  To control for the effect of body size on brain size, we compared the head widths of workers across isolation periods and control groups and found no significant differences (p=0.40, R2=0.098, F=1.18, Z=0.25, SS type 3, Supp Fig 1a).  Our sampling of workers for assignment to isolation or control treatments was thus unlikely to influence brain volume.

Behavioral performance assays

Isolated and control workers used for behavior (n = 16 [2 day], 19 [5 day], 20 [20 day], 11 [40 day], 8 [callow] and 13 [mature]) were exposed to a randomized series of assays to determine foraging ability and ability to recognize and care for brood, tasks that affect colony fitness.  Simplified illustrations of the assays can be found in Fig 1a. 

Predatory Behavior

A single worker was carefully placed in a Fluon®-lined Petri dish (9cm) and allowed to acclimate for 2 minutes prior to assay initiation.  Using Dumont No. 5 fine forceps sanitized with ethanol and dried, a live fruit fly was offered to the worker and the response was observed for 2 minutes.  The degree of predatory behavior was scored on a 5-point scale; higher scores indicate greater predatory behavior.  1. no response or avoidance; 2. olfactory (antennal) investigation; 3. mandibular flaring; 4. latent attack (delayed and intermittent); and 5. immediate attack (persistent predatory attack from first encounter with prey).  The assay arena was also sanitized with ethanol between trials to minimize influence from social signals or prey odors.

Brood recognition and care

To determine if isolates differ in the ability to recognize and care for immature siblings with similar efficacy as control matures and callows, we evaluated their social interactions with pupae.  Each focal worker was carefully placed in a 14 cm diameter Petri dish test arena where a 30 x 10 mm red plastic hollow tube provided a darkened area into which workers could move pupae.  Three pupae were added at three different locations equidistant from the darkened chamber before a worker was introduced.  Workers could relocate pupae either inside or outside of the dark chamber or have no interaction.  Pupae were recorded as aggregated (“piled”) if workers moved two or more together at any location.  Workers showing this level of response began pupal care immediately.  We did not assign lower scores to workers that clustered pupae without moving them into the darkened area.  The darkened tube provided the opportunity for a level of pupal care that would emulate natural placement of immatures within a colony.  However, no worker moved pupae into the tube and we therefore did not include this behavior in the calculation of brood-care scores.  Some workers did not move any pupae, but they themselves moved into the darkened tube.  This was scored as avoidance.

Worker interactions with pupae were observed directly for 5 minutes; after an additional 15 minutes , the final placement of pupae was recorded (total trial length = 20 minutes).  Brood recognition and care was measured on a 7-point scale: 1. no interaction or avoidance throughout the duration of the trial; 2. olfactory (antennal) investigation; 3. mouthpart contact; 4. one pupa moved during the last 15-minute period; 5. one pupa moved in the first five-minute period; 6. two pupae moved beginning in the first five-minute period; 7. three pupae moved beginning in the first five-minute period.  Levels 4 and 5 distinguish the rate of response to pupae, which we interpret as sensory processing ability that could be sensitive to social isolation.  Higher scores indicate greater brood-recognition ability (likely olfactory), restoration of brood clusters through pupal transport, and sustained engagement with brood.  Workers received scores of 5-7 independent of where pupae were transported.

Locomotion

To examine if isolation affected physiological and/or biomechanical capability to move pupae or attack prey, we used a simple and robust proxy to broadly assess neuromuscular impairment.  Focal workers were placed in the center of a 10 cm diameter Fluon®-lined Petri dish divided into four equal  quadrants by a crosshair pattern drawn beneath.  After a two-minute acclimation period, the number of times a worker moved among the four quadrants was recorded during a five-minute period using a digital hand counter.   Movement between quadrants was only scored when the worker passed a division line by at least one body length.

Neuroanatomical and neurochemical measurements
Immunohistochemistry and confocal microscopy

Intact brains (n = 8 [2 day], n = 11 [5 day], n = 15 [20 day], n = 14 [40 day], n = 5 [matures], and n = 7 [callows]) were dissected in ice-cold Ringer’s solution, and then immediately placed in ice-cold zinc-formaldehyde fixative with shaking overnight.  Fixed brains were then washed in HEPES buffered saline 6 times (10 minutes/wash) and fixed in Dent’s Fixative (80% MeOH, 20% DMSO) for 1 hour.  Brains were placed in 100% methanol and stored at -17 C° for 1-3 weeks until processed.  Brains were next rinsed 6 times (10 minutes/rinse) with 3% Triton X-100 in 1X PBS (3% PBST).  Brains were stored overnight in 5% normal goat serum/0.2% PBST/0.005% sodium azide for blocking.  After blocking, brains were incubated for 4 days RT with an anti-synapsin antibody to stain neuropils (1:30, Development Studies Hybridoma Bank, Iowa City, Iowa) in constant nutation.  Brains were washed with 0.2% PBST (6x10 minutes), then incubated in secondary antibody (1:100 AlexaFluor goat ɑ-mouse) in 5% normal goat serum/0.2%PBST/0.005% sodium azide for 4 days at RT with constant nutation.  Brains were washed in 0.2% PBST (6x10 minutes), then dehydrated through an ethanol and PBS series (30, 50, 70, 95, 100% ethanol in 1XPBS).  Brains were cleared and immersed in methyl salicylate and mounted on stainless steel glass-windowed slides for confocal imaging.

Brains were imaged with a Nikon C2 + Si spectral laser scanning confocal microscope (Nikon, Melville, NY, USA) with sections at approximately 1.5 um throughout the entire brain.  Images were manually annotated using Amira 6.0 software to quantify neuropil volumes.  The individual who annotated brains was blind to brain sample category, thus minimizing human bias.  The margins of neuropils were identified visually or with magic wand tools in Amira.  Every eighth frame was annotated manually.  Biomedisa interpolation [48] was used on frames without labels, then evaluated and edited by the annotator for accuracy.  We labeled volumes of the optic lobe (OL; vision), antennal lobe (AL; olfaction), mushroom body (MB; higher-order processing, learning, and memory), lateral calyx (MB-LC), and medial calyx (MB-MC), peduncles (MB-P), central complex (CX; movement and navigation), suboesophageal zone (SEZ; mouthpart control), and rest of the central brain (ROCB).  ROCB describes tissue composed of the superior neuropils, ventromedial neuropils, ventrolateral neuropils, and the lateral horn.  For each brain compartment, volumes of only one intact hemisphere were recorded.  If both hemispheres were intact, the labeler chose one randomly.

A detailed description of brain anatomy and development in relation to age and experience in C. floridanus is presented in Gronenberg et al.  We compared the relative volumes of brains across all groups and scaling relationships among brain compartments.  Relative volumes were calculated by dividing the volume of each focal compartment by the ROCB volume.  Damaged compartments were excluded from analysis.  If one or more compartments or the ROCB was damaged, the sample was excluded.  The MB-LC and MB-MC exhibited similar volumetric trends and thus were combined to one metric (MB-C).

Quantification of brain biogenic amines

Isolated workers were quickly dissected (n = 11 [2 days], n = 10 [5 days], n = 12 [20 days], n = 3 [40 days], along with control mature workers (n = 11) to quantify levels of serotonin (5HT), dopamine (DA), and octopamine (OA) using isocratic, reversed-phase high-performance liquid chromatography with electrochemical detection (HPLC-ED).  Monoamine quantification was performed blind to treatment.  The HPLC-ED system (formerly ESA, Inc., Chelmsford, MA, USA) included a model 584 pump, a model MD-150 (3 x 150 mm) reversed-phase analytical column, a model 5011A dual-channel coulometric analytical cell, and a Coulochem III electrochemical detector.  Brains were dissected in less than 5 minutes in ice-cold Ringer’s solution, homogenized in mobile phase manually with a plastic polypropylene pestle in a centrifuge tube, and centrifuged.  The supernatant was then injected into the HPLC column, and monoamine levels were quantified with reference to external standards.  Mobile phase was formulated as 50 mM citrate/acetate buffer, 1.5 mM sodium dodecyl sulfate, 0.01% triethylamine, and 25% acetonitrile in MilliQ water.  Multiple daily runs of the external standards accounted for any changes in ambient conditions.

Statistical analysis and data visualization

All statistics analyses were performed in RStudio version 4.2.3.  For generalized linear mixed models, we used the package ‘RRPP’.  We tested for statistical significance using ANOVA with residual randomization in a permutation procedure of 1000 iterations and the estimation me­­­­­­­­thod of ordinary least squares, given combined sample size of 25 or more.  We included colony and head width as random effects in our behavior, brain volume, and biogenic amine modeling; neither had a significant effect.  Isolation significantly correlated with head width only when head width was nested with the microscope and experimenter, rather than head width alone, meaning that slight variation between ocular micrometers of the microscopes, rather than treatment, influenced the effect of head width.  ANOVA and modeling details can be found in Supp Tables Set 1 and 2.  Following ANOVA, post-hoc tests were done on pairwise comparisons using Wilcoxon rank sum exact tests with a Benjamini-Hochberg (BH) procedure to correct for multiple tests.  Depending on the sample size of 40-day isolates, we included type III error to account for sample size bias.  Additional nonparametric statistical analyses including Kruskal-Wallis H tests and post-hoc pairwise comparisons using t tests with a Bonferroni correction, which agreed with our modeling, can be found in Supp Tables Set 1 and 2.  We then ran studentized Breusch-Pagan (BP) tests from the ‘lmtest’ package to confirm homoscedasticity in our model.  If heteroscedasticity was found in raw values, we log transformed our data and reran the model and confirmed homoscedasticity. Heteroscedasticity could not be removed from only one model, which was our mixed model analysis of MB-P scaled volumes in Supplementary Tables, Set 1. However, our model nonetheless agrees with our additional nonparametric statistical tests.  We confirmed that log-transformed data resulted in the same pairwise comparison results as raw data, confirming that log transformation for model fitness did not create false pairwise findings.  Because 5-day isolates and 20-day isolates did not have a socially typical, age-matched control and were instead meant to provide insight into the trajectory of neurodevelopment and behavior, we excluded these groups from the analysis reported in the manuscript.  Our analyses including these excluded groups can be found in Supp Figs 1-4 and in Supp Tables, Set 2.   Figures were created using the package ‘ggplot2’.  Figure assembly was done in Adobe Illustrator (Adobe Illustrator 2023).  Illustrations of ants in Visual Abstract, Fig 1a, and Supp Fig 1b are courtesy of BioRender.com.

Excel, RStudio.