Investigating the role of non-helpers in group living thrips

Behavioural variation among individuals is a hallmark of cooperative societies, which commonly contain breeders and non-breeders, helpers and non-helpers. In some cases, labour is divided, with non-breeders “helping”. Conversely, in some societies, subordinate non-breeders may not help. These individuals may be (i) an insurance workforce ensuring continuity of help for breeders when other helpers are lost, (ii) conserving energy while waiting to breed themselves, or (iii) simply of too poor physiological quality either to help or breed. In the Australian Outback, Acacia thrips Dunatothrips aneurae (Thysanoptera) glue Acacia phyllodes into “domiciles” using silk-like secretions, either alone or cooperatively. Domicile maintenance is important for humidity, so repair can be interpreted as helping. I found that not all females helped to repair damage; some repaired partially or not at all ("non-helpers"). At the same time, some co-foundresses are non- or partially reproductive ("non-breeders”), and their role is currently unknown. I first tested the possibility that helping and breeding are divided, with non-helping females breeding, and non-breeders helping. In a lab experiment, I rejected this idea. Experimentally damaged domiciles were typically repaired by reproductive females, and less so by non- or partially reproductive individuals. To test whether non-helpers are an insurance workforce, I successively removed repairing females and found that non-helping females from the same domicile did not increase effort as a result, rejecting this hypothesis. Then I tested whether non-helpers were conserving energy waiting to breed. In a field experiment, I removed all other females, allowing either a helpful female or a non-helper to “inherit” her domicile. Isolated like this, non-helpers laid very few eggs compared to helpers or naturally occurring single foundresses, despite similar ovarian development. My findings show that labour was not divided: reproduction and helping covaried positively, probably depending on individual variation in female quality and intra-domicile competition. Non-helping females were neither an insurance workforce nor conserving energy waiting to breed. They are likely simply of poor quality, freeloading by benefiting from domicile maintenance by others. I hypothesize they are tolerated because of selection for indiscriminate communal brood care in the form of domicile repair.

Methods
Lab and field studies were conducted between January and October 2013, at a large patch of Acacia aneura, the host plant for D. aneurae, at the Bald Hills paddock (S 30° 57′ 39′; E 141° 42′ 18′′) near the UNSW Arid Zone Research Station, Fowlers Gap near Broken Hill, NSW, Australia (see Gilbert and Simpson 2013, Biol. J. Linn. Soc.). D. aneurae domiciles were identified and either collected or marked using coloured tape. I collected domiciles by hand and placed them into ziplock bags inside a cool box, moving them to a laboratory at room temperature within 12 h, and maintained them using protocols described in Gilbert & Simpson (2013). I examined the domiciles under a dissecting microscope. 

Are reproduction and domicile repair divided among foundresses? In domiciles containing two or more females, I made a small tear in the wall (roughly 20% of the silk area), folded back the silk, and observed repair behaviour (defined as stretching strands of "silk" singly between phyllode surfaces) by the foundresses. After scoring in this way, all group members were removed and dissected under a microscope (see protocol below) for assessment of ovarian status. I also measured female body size to the nearest 20 µm (pronotum width at 50⨉ magnification).

Are non-helping foundresses an insurance workforce for breeders? In a subset of 19 domiciles, I counted the individual silk strands applied by each female while repairing the experimental damage. After 1 h I removed the first repairer to respond and observed the behaviour of the remaining female(s) over the subsequent 1 h period. Over successive 1 h periods, I removed one-by-one the repairers in the sequence that I saw them begin repairing and counted silk strands laid down by the remaining females. Removed individuals were dissected and scored for ovarian development as described below, and their pronotum width was measured.

Are non-helping foundresses waiting to breed themselves? In a field removal experiment using domiciles containing 2 or 3 females, I first categorised each female as “repairer” or "nonrepairer", as above: I slightly damaged the domicile wall, then observed the behaviour of the inhabitants for 3 h. I used a longer observation period of 3 h under field conditions where observation was more difficult, to be sure to identify nonrepairing females. After classifying, I removed all individuals from all domiciles. I then assigned domiciles randomly to treatment groups in which I reinstated to her original domicile either a repairer (R group, n=11) or a nonrepairer (NR group, n=8). In this way, I allowed (or forced) this female to “inherit” her domicile. Replacement was completed within 30 minutes of removal. All other females I dissected to assess reproductive status as described below. The remaining female I left in the domicile. I also created two control groups (CS, control singletons, and CP, control pairs). The CS group consisted of naturally occurring singly built domiciles (n=46), while the CP group consisted of domiciles naturally built by pairs of females (n=9). In the CP and CS groups, I performed the same experimental domicile damage before removing and immediately replacing all domicile inhabitants. After manipulation, I monitored domiciles for 21 days to assess (1) the time to repair the domicile, and (2) the number of offspring produced. 

Time to repair domicile: Each day I visually categorised repair as “no repair”, “incomplete repair” or “complete repair”. “No repair” was where a negligible number of silk strands had been added and the damaged area remained substantially open to the environment. “Incomplete repair” was where repair had begun, but strands were sparse and perturbation of the phyllodes (e.g. by mild wind) would likely reopen the hole in the silk. “Complete repair” was where silk completely covered the damaged area, perturbation of the phyllodes would be unlikely to reopen the hole in the domicile wall, and/or the repaired wall resembled the original silk. I also recorded any instances of naturally occurring domicile damage (and its repair), destruction, or abandonment. 

Offspring in domicile: After 21 days, I counted the number of offspring (eggs and larvae, live or dead) in each domicile, and dissected the female to examine her ovaries (see below).  All individuals remaining in domiciles after the 21 d experimental period were also dissected as described below to assess reproductive status. 

Assessing ovary status: I assessed ovarian development in all dissected individuals following Gilbert et al (2018, Scientific Reports). Foundresses were killed by immersing in 100% ethanol for 1 m and were dissected immediately in water. I recorded the number and volume to the nearest 20 µm at 50⨉ (π × length × (width/2)2), assuming eggs are an ellipsoid) of any developing oocytes, and also of any mature, chorionated eggs that were ready to lay, and summed the resulting volumes to give a total volume within each ovary. For analysis, oocyte and mature egg volumes were log-transformed after adding the minimum value in the dataset to all values.