Fecal odors attract cockroaches to aggregations, where conspecific feces provide nutrients and seed the gut microbiota. Because these odors arise from the gut and fecal microbial communities, which are shaped by coprophagy and diet, discrete aggregations within a heterogeneous landscape can produce distinct fecal odor profiles. How such odor diversity influences recognition, group-affiliation and ultimately aggregation remains unclear. We manipulated coprophagy and then diet quality in gnotobiotic nymphs to generate groups with distinct fecal odor signatures, then raised them to the adult stage and collected their feces for behavioral assays. Naïve nymphs lacking coprophagy experience were equally attracted to fecal odors from all treatment groups, suggesting an innate attraction to conspecific feces. In contrast, experienced nymphs preferred the fecal odor of their natal group over foreign groups, suggesting learned group-specific recognition. In associative learning assays pairing a fecal odor with a glucose reward, nymphs preferentially aggregated with the conditioned odor over novel odors. These results show that aggregation in neonate cockroaches is mediated by both an innate attraction to conspecific fecal odor and learned preferences for specific odorants encountered during coprophagy soon after they hatch. We propose that these mechanisms guide neonates toward natal aggregations, facilitating acquisition of a locally adapted gut microbial community.

METHODOLOGICAL INFORMATION

1. Description of methods used for collection/generation of data:

Materials and Methods

(a) Insects

The laboratory colony of B. germanica (Orlando Normal strain, also known as American Cyanamid, collected in a Florida apartment in 1947) used in these experiments was reared on rodent diet (Purina 5001 Rodent Diet, PMI Nutrition International, St. Louis, MO, USA) and provided glass-distilled water in cotton-stoppered vials. The main colony and all cockroaches used in this study were maintained at 27 ± 1 °C, 40–70 % relative humidity, and a 12-h L : 12-h D photoperiod.

(b) Gnotobiotic first instar nymphs with axenic guts

The German cockroach forms an ootheca (egg case) that houses ~40–45 fertilized eggs. The female carries the ootheca for a ~20-day embryonic development. Embryos that are close to hatching develop pigmented eyes and a prominent green spot representing yolk in the hindgut. When a “green line” was visible in the ootheca, the gravid female was briefly anesthetized with CO2 and the ootheca was gently removed with forceps. We sterilized the exterior surface of oothecae by washing them for 1 min in a 0.5 % bleach-water solution, then 70 % ethanol in water for 1 min, and three times with water for 1 min each. All solutions were made with autoclaved water. To determine if the sterilization process was effective to generate axenic first instar nymphs (henceforth, first instars), two surface-sterilized oothecae were plated on tryptic soy agar (TSA) and plate count agar (PCA). After 3 days of incubation at 28 °C, no colonies were visible in any of the treatments. This surface sterilization procedure resulted in gnotobiotic neonates that lacked a gut microbiome, but they retained their maternally provisioned endosymbionts, namely Blattabacterium. These gnotobiotic nymphs could be maintained on TSA plates for several weeks without any bacteria growth (data not shown), confirming that their guts were axenic.  

To generate the experimental groups (200 nymphs/treatment), 20–30 surface-sterilized oothecae were placed in a sterile 15 ml conical centrifuge tube (Falcon-Corning, Corning, NY, USA) and kept in a rearing cage (18.7 x 13.3 x 9.5 cm, T-79, Althor Products, Windsor Locks, CT, USA) that had been sterilized with bleach and 95 % ethanol. To maintain moisture, sterilized glass test tubes containing autoclaved water were placed in the cage until the nymphs hatched. After hatching, the gnotobiotic first instars were randomly assigned to treatment groups.

(c) Three types of diets

To obtain cockroach feces of different qualities and odor profiles for the aggregation bioassays, we used three non-sterile diets with variable protein, carbohydrate, lipid, and fiber contents: Rodent diet was the standard diet for our colony and it contained moderate protein (24 %), high carbohydrate (58 %), and low lipids (5 %) and fiber (5.2 %); Cat diet (Taste of the Wild Rocky Mountain Feline Recipe with roasted venison and smoked salmon, Schell & Kampeter Inc., Meta, MO, USA) contained high protein  (42%), low carbohydrate (21 %), moderate lipids (18 %) and low fiber (3 %); Rabbit diet (Oxbow Garden Select Adult Rabbit Food, Oxbow Enterprises Inc., Murdock, NE, USA) had low protein (12%), high carbohydrate (4 3%), low lipids (2.5 %) and high fiber (22–26 %). Each diet was crushed in a blender and placed into separate Ziploc bags. Sterilized rodent diet was prepared using gamma radiation at a dose of 10,000 Gy at the USDA-APHIS-PPQ S&T Otis Laboratory (Buzzards Bay, MA, USA). To verify the sterilization procedure, two samples of the irradiated diet were plated on TSA and PCA media using 4 plates per sample. After 3 days of incubation at 28 °C, no colonies or visible growth were detected in any of the diet and control plates.

(d) Generating three treatment groups

To test the effects of bacteria sources and diet on aggregation preferences to fecal extracts, we generated three experimental groups. General procedures common to all three treatment groups were as follows. Gnotobiotic first instars (with axenic guts) were placed in a sterile rearing cage (100 nymphs/cage). Each cage contained an autoclaved egg carton shelter and autoclaved water in a sterilized glass test tube (25 x 150 mm, Fisherbrand, Fisher Scientific Company, Hanover Park, IL, USA). Each of the three experimental groups had two replicate cages. Routine maintenance every two weeks was conducted within a biosafety cabinet. The nymphs and their shelter were transferred without anesthesia to a new rearing cage to maintain a clean, sterile environment. Water and diets were replaced as needed. The first date that nymphs molted to the next development stage in each cage was recorded. The date of each adult emergence, the sex and body mass were recorded daily.

To obtain fecal materials and first instars for bioassays, newly emerged adult males and females from each treatment group were transferred daily into sterile mating cages with sterile shelter, diet and water. Gravid females with a “green line” visible in their oothecae were then transferred to a sterile hatching cage with sterile shelter and water. Newly hatched gnotobiotic nymphs were used in bioassays. Thus, the adults were considered the end of the parental generation, and the offspring nymphs they produced were used in bioassays.

To collect fresh feces from adult females, 20–50 females carrying oothecae were collected. We carefully dislodged their oothecae, which stimulated the females to initiate a new gonotrophic cycle of feeding, oocyte provisioning, and feces excretion. The females, along with a sterile shelter, diet and water were placed in a sterile cage with a screen (12 Mesh T316 Stainless, 0.58 mm wire diameter, 1.53 mm opening, TWP, Berkeley, CA, USA) at the bottom. We collected the feces that passed through the screen during one week. The diet was placed in a small Petri dish to prevent it from falling through the screen, but any diet debris were carefully removed from the feces under a microscope.

The three treatment groups were defined as follows:

Different-Diets (no coprophagy). To evaluate the effect of diet on the development of gnotobiotic nymphs, one-day-old first instars with axenic guts received sterile water and one of three non-sterile diets (1 g of rodent, cat or rabbit diet, hereafter denoted D-Rodent, D-Cat and D-Rabbit, with “D” representing “Diet”) in a Petri dish (60 mm diameter, Falcon-Corning, Glendale, AZ, USA); they were not inoculated with fecal bacteria (no coprophagy). We expected their fecal odors to differentiate as their gut microbiomes diverged on the different diets and associated environmental bacteria during nymphal development.

Coprophagy & Different-Diets. To evaluate the effect of coprophagy on the development of gnotobiotic nymphs, one-day-old neonates with axenic guts received sterile water and non-sterile feces (0.5 g on a 60 mm Petri dish) excreted by virgin females from the laboratory colony; these females were maintained on non-sterile rodent diet. The nymphs were given 3 days to inoculate their guts with fecal bacteria. Then, the Petri dishes with feces were removed and groups of nymphs received one of three different non-sterile diets as in the “D” treatment above (hereafter denoted FD-Rodent, FD-Cat and FD-Rabbit, with “F” representing “Feces”). We expected the ingested fecal bacteria to inoculate all nymphs with a common gut microbiota, which would then be differentiated by the different diets and environmental bacteria during nymphal development, producing unique fecal odor profiles.

Coprophagy & sterilized-Rodent-diet. We expected cockroach groups raised on different diets to differ in their gut and fecal microbiomes and therefore in their fecal odors. To test whether inoculation with different fecal bacteria via coprophagy would impact the development and olfactory preferences of nymphs fed the same diet, one-day-old gnotobiotic neonates received sterile water and three different types of feces (from females in the FD-Rodent, FD-Cat or FD-Rabbit groups) for 3 days (0.5 g on a 60 mm Petri dish). After removing the Petri dishes with fecal inoculum, the nymphs were offered irradiated rodent diet. Hereafter, these treatments are denoted RoFD-stRodent, CaFD-stRodent and RaFD-stRodent, with RoF, CaF, and RaF representing rodent, cat, and rabbit feces, respectively, and stRodent representing sterilized rodent diet). Previous studies showed that diet quality strongly influenced and shaped gut bacterial communities. Therefore, we expected that gnotobiotic neonates inoculated with different bacteria from different fecal sources would develop divergent gut microbiota. At the same time, we expected that the common sterile diet throughout nymphal development and during the adult stage would converge digestion pathways and the gut environment and thus unify the gut bacterial communities of the three treatments.

We also set up a single cage containing 100 nymphs to monitor development of gnotobiotic nymphs after a brief starvation period that mimicked the exposure to feces, followed by exposure to sterilized diet throughout nymphal development. These gnotobiotic neonates did not receive feces for one day (starved, no coprophagy) and then received irradiated rodent diet (No-coprophagy & sterilized-Rodent-diet).

(e) Bioassays using first instars
Feces were collected from three groups of adult females fed the three non-sterile diets (i.e., rodent, cat or rabbit diets). The feces were weighed and homogenized (50 mg feces/ml sterile water) in a 1.5 ml Eppendorf tube. We vortexed the suspension for 3 min, then centrifuged (14,000 rpm for 1 min at 4 °C) and collected the supernatant in a 1.5-ml sterile Eppendorf tube. We defined the concentration of this initial extract as “1X” (50 μg feces-equivalents/μl).

1. Dose-response assays. Feces originating from the three different treatments varied in their effectiveness at stimulating aggregation. To standardize the stimuli used in two-choice aggregation assays, we estimated the Effective Concentration (EC) of each fecal extract using first instars that hatched from the respective diet treatment (i.e., offspring nymphs). Each of the fecal extracts from adult females fed different diets was diluted 10-fold (0.1X = 5 μg feces-equivalents/μl) and 100-fold (0.01X = 0.5 μg feces-equivalents/μl) with sterile water. Two accordion-style folded shelters (20 x 6.7 mm, Whatman #1, Pittsburgh, PA, USA) were placed at opposite edges of a Petri dish (60 mm). One shelter was treated with 1 μl of the fecal extract of the tested nymph’s own treatment group. The other shelter was treated as control with 1 μl of sterile water. A single first instar was allowed to walk onto an autoclaved wooden stick (140 x 5 mm, Eisco™ Wooden Splints, Fisherbrand) and guided to descend onto the center of the arena; thus no anesthesia was used. Bioassays were initiated during the photophase, when cockroaches normally aggregate, and the position of the nymph within the arena was recorded 2 hrs later. We conducted 20 to 60 assays per concentration. The EC50 (effective concentration) and EC70 were calculated from the resulting dose-response curve for each group. Based on these results, we used 3 μl of 5 μg feces-equivalents/μl fecal extract for all subsequent bioassays.

2. Two-choice aggregation assays. The experimental aggregation assays were identical to the dose-response assays, except we used 100 x 15 mm Petri dish arenas (Fisherbrand) and tent-shaped shelters (20 x 20 mm, Whatman #1). Based on the dose-response bioassays, Shelter 1 and Shelter 2 were treated with 3 μl of sterile water or fecal extract (5 μg feces-equivalents/μl). The overall response of nymphs was calculated as Response (%) = (100 × (# of nymphs choosing shelter 1 and shelter 2 / total # of tested nymphs)). A preference index was calculated as Preference (%) = (100 × (# of nymphs choosing either shelter 1 or shelter 2 / total # of nymphs that chose shelter 1 and shelter 2).

(f) Olfactory associative learning
Using gnotobiotic neonates, obtained by surface-sterilizing the oothecae of lab-colony females fed rodent diet, we generated five experimental groups. Newly hatched groups of approximately 200 gnotobiotic neonates with axenic guts were placed on a sterile filter paper strip (20 x 80 mm, Whatman #1) that served as a shelter in a sterile 50 ml conical centrifuge tube (Falcon-Corning, Corning, NY, USA) with sterile water in a 1.5 ml Eppendorf tube.

1. No-conditioning (N). Nymphs did not receive any conditioning.

2. Coprophagy-conditioning (F) (Fecal nutrient and fecal odors). To mimic natural coprophagy, nymphs received sterile water in a 1.5 ml Eppendorf tube and 0.2 g of non-sterile feces collected from females fed non-sterile rodent diet (FD-Rodent) moistened with 10 µl sterile water. In this paradigm, nymphs engaged in coprophagy while being exposed to fecal odor for either 1 day or 3 days.

3. Glucose-agar-conditioning (Glu) (Glucose nutrient only). Nymphs received 2 ml of a blue-colored glucose-agar mix (1 mol L–1 glucose, 1 % agar and 0.5 mmol L–1 erioglaucine). Erioglaucine, also known as Brilliant blue FCF, is a food-grade dye that is not toxic to cockroaches; it was used to confirm that the nymphs consumed the glucose-agar mix. Nymphs were exposed to the glucose-agar for either 1 day or 3 days but not to fecal odor other than their own feces.

4. Fecal odor-conditioning (FO) (Fecal odor only). Nymphs received fecal odor from 0.2 g of FD-Rodent female feces moistened with 10 µl sterile water. Feces were placed in a 1.5 ml Eppendorf tube capped with a piece of cheesecloth (Monarch, Knoxville, TN, USA). Thus, nymphs were exposed to fecal odor but prevented from contacting the feces for either 1 day or 3 days.

5. Glucose-agar & Fecal odor-conditioning (Glu+FO). Nymphs received 2 ml of blue-colored glucose-agar mix, as above, while being exposed to fecal odor, as above. Thus, nymphs were prevented from contacting the feces for either 1 day or 3 days but were exposed to fecal odor while eating glucose-agar.

Fecal odor preferences of conditioned 1-day-old nymphs (1 day conditioning) and 3-day-old nymphs (3 days conditioning) were tested in two-choice aggregation assays using the fecal extracts of FD-Rodent, FD-Cat and FD-Rabbit (3 µl/shelter, 5 μg feces-equivalents/μl). We expected that nymphs in Experiment 1 (No conditioning) would be equally attracted to different fecal extracts, while nymphs in Experiment 2 (Coprophagy-conditioning) would prefer the FD-Rodent fecal odors over other fecal odors (FD-Cat and FD-Rabbit). If this fecal odor preference were due to associative olfactory learning during feeding, nymphs in Experiments 3 (Glucose nutrient only) and 4 (Fecal odor only) should not have a specific preference for FD-Rodent fecal odor. On the other hand, in Experiment 5 (Glucose-agar & Fecal odor-conditioning) nymphs were expected to prefer FD-Rodent fecal odors. The preference index was calculated using 30 replications of each assay.

(g) Data analysis

We found no differences in the nymphal development periods between the two replicates within each treatment group. Therefore, we combined both replicates for the analysis of nymphal development times (adult emergence) and the bioassays that followed. The effects of diet and coprophagy on nymphal development and adult body mass (parental generation) of the Different-Diets group and Coprophagy & Different-Diets group were tested by two-way ANOVA (α = 0.05) followed by Tukey’s HSD, which enabled multiple comparisons of main (fixed) effects including development with vs. without coprophagy (Coprophagy main effect) across the three diets (Diet effect) and the interaction of these two factors (Coprophagy * Diet). Odor preferences of first instars (offspring generation) were assessed using a Chi-square test of independence (α = 0.05). All statistical analyses were performed in JMP (Student edition 18, Cary, NC, USA). Details of all statistical tests are shown in SI.

2. Methods for processing the data:

The effects of diet and coprophagy on nymphal development and adult body mass (parental generation) of the Different-Diets group and Coprophagy & Different-Diets group were tested by two-way ANOVA (α = 0.05) followed by Tukey’s HSD, which enabled multiple comparisons of main (fixed) effects including development with vs. without coprophagy (Coprophagy main effect) across the three diets (Diet effect) and the interaction of these two factors (Coprophagy * Diet). Odor preferences of first instars (offspring generation) were assessed using a Chi-square test of independence (α = 0.05).

3. Instrument- or software-specific information (including package versions) needed to interpret the data:

All statistical analyses were performed in JMP (Student edition 18, Cary, NC, USA).

attached base packages:
N/A

Changes after Dec 5, 2025: Figure4_DRYAD_Group_recognition_in_a_subsocial_extreme_omnivore_is_based_on_fecal_odor_preference_that_is_modulated_by_coprophagy_diet_and_learning.csv was replaced, and README_DRYAD_Group_recognition_in_a_subsocial_extreme_omnivore_is_based_on_fecal_odor_preference_that_is_modulated_by_coprophagy_diet_and_learning.csv was removed in Jun 23, 2026.