24/3723, 28:03 Who needs lazy workers? Inactive workers acl as a reserve’ labor force replacing active workers, but inactive workers are not. Who needs ‘lazy’ workers? Inactive workers act as a ‘reserve’ labor force replacing active workers, but inactive workers are not replaced when they are removed tie Selontrs 217+ ng e087iomalpoe18ar« Abstract Social insect colonies are highly successful, soltorganized complex systems. Surprisingly however, mast social insect colonies Contain large numbers of highly inactive workers, Although this may seem inefficient, may be that Inactive workers actually Contribute fa colony function. Indeed, the mast commonly proposed explanation for inactive workers 's that they form a Taserve! labor force that Becomes active when needed, thus helping mitgate the effects of colony workload fuctuations or worker loss. Thus, it may be that inactive workers facitate colony flexibity and reslience. However, ths idea has not been empirically confirmed. Here we test whether colonies of Ternothorax rugatuus ans replace highly acve (spending large proportion of time fon specific tasks) or hight inactive (spending large proportions of time completely immobile) workers when they are experimentally removed. We show that colonies maintained pre-femavalactviy levels even after acive workers were removed, and that previously inacve workers became active subsequent fo the removal of active workers. Conversely, when inactive workers were Femoved, inactvy levels decreased and remained lower post-emoval. Thus, colonies seem fo have mechan sms for maintaining @ Certain number of active workers, but nota set numberof active workers. The rapid replacement (within 1 week) of active workers suggests tat the tas they perform, mainly foraging and brood care, are necessary for colony function on short imescales. Conversely, the lack of replacement of inactive workers even 2 weeks after their removal suggests that any potential functions they have, including being a ‘reserve are ess important, or auxiliary, and do not need immediate recovery. Thus, inactive workers act 18 a reserve labor force and may sil playa role as food stores forthe colony, but a role in acl tating colony-wide communication is Unlikely. Our results are consistent with the often clad, but never yet empirically supported hypothesis that inacive workers act as a ool of reserve" labor that may allow colanies to quickly take advantage of novel resources and to mitigate worker loss, Citation: Charbonneau D, Sasaki T, Dormhaus A (2017) Who needs ‘lazy’ workers? Inactive workers act as a reserve’ labor force replacing active workers, but nacive workers are not replaced when they are removed. PLOS ONE 12(9) @0184074. hitpsidoiorg/10.137 Vjoumal pone. 0184074 Editor: James A. R. Marshall, University of Sheffield, UNITED KINGDOM Received: September 1, 2016; Accepted: August 17, 2017; Published: September 5, 2017 ‘Copyright: © 2017 Charbonneau etal. This I an open access article distributed under the terms of the ‘Alinbution License, which permits unrestricted use, istibution, and reproduction in any medium, provided the original author land source are credte. Data Availabilty: All data fles are avalabe ftom the Dryad database (Provisional DOI: dol 10.606 Veryad, 77110), Funding: Ressarch supported through the GIDP-EIS and EE8 Department at University of Arizona, as well as NSF grants no. 1OS-1048238, 1OS-0841756, and DBI-1262282 (te AD). The funders had no role study design, data collection ana analysis, decision to publish, or prepaation of the manuscript ‘Competing interests: The autnors have declared that no campatng interests exist. Introduction Complex systems are a broad class of systems in which behavior emerges from the actions and interactions of a number of independent units. Those can range from human-made systoms such as computer networks [1.2], rabol swarms [3.4 transporation networks [5.4], human organizations [7], and economic systems [i]. a8 wel as biological systems such as ‘embryogenesis and organogenesis [9], cisease transmission networks [10], genes expression networks [11~13}, the organization of multcelula systems [4), and social insect colonies [16-17]. In many cases, complex systoms are decontralzed, sel-organized land optimized for group-level function ‘Social insect colonies are highly successful, evolved, self-organized collectives which are oftan used as models fr the organization of complex systems. They are thought to employ sophisticated indvidual and group-evel svatogies forthe alocation of workers to tasks. For example, incvidual honey bee workers adjust ther foraging strategy accorcng fo innate processes, leamed information, and social signals [18-21], such that group level decisions will depend on the amount of stored food [22-24 the rate of food ‘consumption [25), and the food availabilty in the environment (26). Thus, task allocation in insect colonies is expected tobe flexible to changes In demand for diferent types of work [2]. and robust to individual failure [28 28). ‘Studies show that experimentally increasing workload, such as by increasing the temperature in honey bee hives [30.34] or inreasing food avalabilty [32-34], leads fo increased activiy and recrutment of new workers to underserved tasks. Additionally, removal studies tha simulate worker loss (by death or atherwse) show tha the lost workforce tends tobe replaced by other ‘workers in the colony [30,35-40],Iis worth noting that, although social insect colonies are generally thought to be highly flexible and robust, thar is also evidence that ths isnot always the ease and that rapid worker reallocation does nol always eccur when conditions change (ether by fluctuating workload [41,42] or by worker loss (43.44), hitps:iournas plos.org/plosonelartila?id=10.137 journal pone.0184074 ana24/3723, 28:03 Who needs lazy workers? Inactive workers acl as a reserve’ labor force replacing active workers, but inactive workers are not. Despite these sophisticated mechanisms of task allocation, perhaps one of the most surprising features of socal insects is that high levels of inactivity are commen in most species, Socal insect colonies typically have upwards of 50% oftheir workers inactive at any on time (honey bees [15,28 46}, bumble bees [47, wasps (48), termites [49], and ans (50-54). Athough individual activity level may vary over the course ofthe day, or across days, the relative actly ranks of individual workers are consistent among, ‘workers over moderate ‘imescales (one toa few weeks [35,35.50,55-S7), indicating that workers are consistently more or ess active AAlhough the abundance of highly inactive workers may seem counterintuitive given the sophisticaton and flexibly of socal insect task allocation strategies, it may be that inactive workers actually contribute to colony function rather than impairt indeed, the ‘mast commonly cted explanation fr inactivity proposes that inactive workers are eserves that serve as an auxllary labor force that can help the colony react quickly lo workload increases (otter proposed hypotheses reviewed in [17 58). Thus, ifinactive ‘workers constitute a pool of workers that can be dynamically allocated to tasks and therefore adjust the amount of work performed by the colony to fluctuating demands, their presence may allow colonies to be increasingly flexible and reset (see [17] for an extended discussion of ths hypothesis). ‘Studies that experimentally remove workers show that colonies tend to replace lost workers. However, these studies typcaly oly remove key workers such as bees hat inerease air movement in colonies by fanning (SS), workers acive mn emigration tasks such 23s Scouting new nest locations a recruiting new workers to potential nest locations (SZ), workers that remove ant corpses from the est undertakers) [50], nurses that care for brood [Bt and foragers [62.63] In these cases where workers are identified in only, tne task, the individuals remaved tend tobe highly active workers. Inrmast cases cited above, removed acve workers are eectvely replaced: their tasks and workloads are taken over by cther workers and performed at comparable actvty levels (though 500 (43,44), Replacement can be immediate (e.g, water foragers [62)), but there is typically a delay ranging fom a few hours toa few days (57,60-8 In studies where pre-removal behavior was known, replacement workers tended to be less active, but not inactive, workers previously performing the replacement task ata lowe’ level, who inereased ther activity levels to compansate fr the lose of highly Active workers (61.62.84) Thus, although the ‘reserve’ worker hypothesis s commonly proposed, this isnot direct evidence for Since actually inactive’ workers were nat involved. The few studies thal clam to support the reserve worker hypathes (35,65) define nactvty asthe absence of observation ofa set of pre-determined behaviors, Consequently, workers defined as acives! may infact have been performing a wide range of tasks (e.9.patroling, grooming, and bulking) that were not par ofthe pre- Setermined task lt, and so may not have been truly inactive at al Infact, the only study that has successfull ‘activated’ inactive ‘workers removed al other workers excep! forthe inactves [5]. There i theoretical evidence that such workers could ensure Colony survival in the case of a major catastrophe that eliminates o fatigues all ther workers (82) Inactive workers represent a signicant investment of colony resources, typically making up more than half of all workers. This suggests the possibilty that they play an important yet unknown role wihin the Colony, in which case we should expect them to be Feplacediflost. Many potential lunctons olher than a reserves’ have been proposed to explain the presence of highly inactve ‘workers (reviewed in [17] and [58). These include inactivity as a form of social ‘cheating in which eqg-aying workers selfishly investin their own reproduction rather than contrbute lo colony flness by avoiding risky tasks and conserving energy by remaining inactive (tested and supported [6869]; tested, but not supported (70.52.68), and inactive workers performing an as-yet unidentified function, suen as playing a ole in communication (proposed in [74 tested, but not supported [58) and acting as food stores, or repletes (tested and supported [58 72). Thus, f inactive workers have a function other than as reserves, we expect the calony to have mechan'sms to engure that workers are allocated lo thal function just as we see for active workers, In this paper, we test whether socal Insect colonies have mechanisms thal maintain colony activi levels homeostaticaly (inthe broad sense of retuming to a desired value ater perturation) by removing ether tre 20% most active, the 20% most inactive, or the same proportion of random workers, then racking the actully levels of the remaining warkers pastremoval Methods We collected 20 colonies of Temnothorax ragatulus ants (1307 workers total, colony size mean = 86.35, median = 88, sd. = 93.04) from a pine forest located at ~2000f in allude inthe Sana Catalina Mountains, USA (N32.395 W110.688) in June and July of 2018 (15 colons) and 2016 (5 colonies; soe S1 Tabla). The colons were collected ina National Forest (Saguaro Natonal Park East) which does not require a peril for callecions o this scale. The collection consistec of 20 colonies of Temnothorax rugaius ants, an abundant and widely disinbuted species that is neithor endangored nor protected Within 1-2 days of ther collection, colonies wore allowed to emigrate to artical nests consisting of apiece of cardboard with an fenclosed nest area and enfrance die-cut out ofthe middle, sandwiched between ‘wa glass slides [50 Colonies were Kept in he lab ‘ona 12h12h light regime (Bam to 8pm) and fed water, diluted honey solution, and wingless ft tes ad fibtum (similar methods to (say. To test the effet of ‘active’ and ‘inactive worker removals, we applied one of three treatments to colonies: removal ofthe 20% most active workers (§ colonies), romoval ofthe 20% most inactive workers (8 colonies), and removal of 20% randomly soloctod workers {G colons; see below for detailed doscripions of worker actvty and inactivity measurements). From this point on, we wil refer to the 20% most active and 20% most inactive as ‘active’ and ‘inacve' in quotations). Whole calories wore fled over 3 consecutive {days before removals (shown to be sufficient! fo abtain consistent indivdual behavior (50), at one week post-ramavals, and at two ‘woeks post-removal (total af 8 videos per colony). Typical liming of events was as flows: Day 1: colonies calocted and re-housed, Day 4: workers painted, Oay 6-8: pre-removal videos, Day 8; workers removed, Day 13-15: 1 week post-removal videos, Day 21— 28:2 weoks post-ramoval videos. There wore occasionally delays of up to 1 week Between collation ane panting, but the schedule was consistently maintained once fling had begun, Each video was 5 minutes long. Colonies were flmed using an HD camera (Nikon D7000 with Nikon AF-S Micro-NIKKOR 60mm £72.86 ED Lans and Lumix DMC-GHS with Olympus OL6028 6mm #2.8 Macro Lens) mounted direc aver the nest with three sources of dfused ight to reduce shadows. The entire nest, as wel as the avallable food and water outside ofthe nest, were within the fale of view. All videos were taken within ~1 month of colony collection to imit potential laboratory elec, such as artificial age situctures due to inereasod foragor ago, Previous work has shawn that colony time spent on any specie task (ineluging inactivity) o net signifcantly vary between Feld ard laboratory within tese timeframes (73) hitps:iournals plos.org/plosonelartila?id=10.137 journal pone.0184074 ana24/3723, 28:03 Who needs lazy workers? Inactive workers acl as a reserve’ labor force replacing active workers, but inactive workers are not. Within t weok of colony emigration to atfcial nests, each worker was painted with a unique color combination to alow individual tracking between observations (see [80] for detailed metnods). Worker mortalty from paint marking (typically resulting from painting ‘ver antennae, leg joints, or spiracles) was estmated to be faily low (<5%), but was not systematically trackad, Colonies were fimed at least 48h after workers had been painted to allow colonies to retumn to their normal function, Videos were analyzed by tracking each indlvidual worker over the course of the 5 minute video and recording the task the worker _was pesforming for each second ofthe video (a Ist of possible tasks can be found in Table 1). Tasks were classed as ether ‘active buléing, foraging, brood care, sel- or allo-grooming, rophaliaxis, and feeding on dead frat les that had been brought back fo the nest), undiferentiated’ (walking inside the nest and not otherwise engaged in any active task), or ‘inactive’ (mmaDie {and not otherwise engaged in any active task) actly, code, and detailed docrptions. For every second of analyzed video, each ant has one of these behaviors attrbuted to. [Smalar to 72). ‘nosuidol org/10.137'1/ournal pone 0184074001 Worker actvty and inactivity levels were calculated by averaging the proportion of observed time spent an active tasks or inactive, respectively, for videos from each black of 3 consecutive days ofiming (pre-remaval, 1 week postramoval, and 2 weeks post Femaval). Inactity a8 defined here i rot simply the Inverse of atv (0. the absence of acti) because worker time Is Sistriouted among time spent active, inactive, and wandering inside (ve. undiferentiated activity, which isnot counted towards tither activity or inactivity). Workers that wore observed only once during each block (appeared In only one of the three videos) ‘were removed from the analyses fo ensure an adequate representation of worker ime budgets. OF the workers with suficent data, the 20% with the highost activity and inactivity levels were removed forthe ‘active! and ‘inactive removals spectively. We selected removals of 20% of workers as this should be sufficient to cause a significant effect on colony actviyfinactviy, while ensuring @ range of activiyinacvity els in the remaining workers, Random removals were selected using the R function ‘sample\) Previous work has shown that workers of Tomnothorax rugatulus ants can be grouped into four distinct task groups, or behavioral castes: inactive, foragers, nurses, and walkers [50.58]. These groups specialize (spend more time relative to other workers) an inactivity, foraging and building, brood care, and wandering inside respectively (Table 1). Here, we use these tasks groups to determine the roe of workers within calanies prior to removals. Using the mean proportion of ime spent on tasks pre-removal, we established pre-removal worker task groups using a combination cof principal eomponent and hierarchical cluster analyses (prcomp and hiclust, base ‘sats’ package in R Version 3.1.2) as follows. We determined the numberof cistinct task groups (4 by dentying tasks that conte the most tothe principal components (tasks with absolute sum > 1; rumber of components determined via parallel analysis [74], We then Used a tierarchical clustering Analysis to classify workers into distinct task groups based on similares in ther ime spent on tasks (worker custerng i identical to BB}; see St Fig). Workers wore clustered into 4 distinc task groups: 1) Inactves (605 workers~ 40.8%), (2) Foragers (145 workers 9.820), (3) Nurses (249 workere~ 16.8%), and (8) Wakere (483 workers 32.5%) (task group names capitalized from this ont on). The Inactive task group, abtaines via hierarchical clustering analysi, should na to be confused with the “nactve’ group, Which represents the 20% most inactive workers as determined by the continuous variable of %& time spent inactive. in order to ‘ccount for intercolony varation in nactvly levels, the data foreach colony were centered (task mean subiracted from task Values) and scaled (task values divided by task standard deviation) separately before being pooled and clustered. To avoid the effects of sporadic low repeatability tasks (such as grooming and eating), we performed the clustering analysis only on highly consistent (high repeatabilty) tasks, namely inactivity, wandering inside, foraging, bood care. The same groups (Foragers. Nurses, Walkers (her patlles), and Inactves were obtained previously va clustering analysis that included eae individual task (50) Removing les repoalabe tasks from the analysis allows for clustering thats less dependent on small random variation n tasks that are not representative of worker specialization. These methads have been used to deserbe colony organization in prior studies [28.28 incluaing on Temnothorax rugatulus [30.581 ‘Statistical analyses were performed in R (Version 3.0.3), and consisted of Mixed-flects madels and Tukey post-hoc tests (packages 'nime’v3.1-115 and multcomp'v1.3-2), as well as Wilcoxon Signed-Rank tests (base ‘stats’ package, wlcoxtest fUneton). Mode formulae can be found inthe figure legends, and generally include the predicted fixed effect (ypicaly mean individual worker sctvtyinactiviy-.e, average amount af time sper actve/inactve by an individual relative to fatal observation time) a well as colony ideniy as a random effect. The fixed effect ‘Tra elers to ime powtin the experiment (pre, 1 week post, 2 weeks post), while ieatment refers fo the removal treatment (active, inactive, random), Results ‘The across-colony mean proportion of time workers spent on active Lasks was 0.170 (median = 0.154, . = 0.071). The mean proparton ofime spent inactive was 0.607 (median = 0,628, 8.6, = 0.146) (see $2 Eig for distribution of colony actly levels) ‘which is consistent with previous work on Temnethorax rugatlus (50,5873) To ensure that worker removals should have a significant impact on colony actly and inactivity levels, we compared the activity and inactivity levels of whole colones (Le. mean activty levels of ll workers n a colony) pre-emoval ta the calculated mean colony ‘activity level of colonies without the workers that would later be removed (Le. mean activity levels of all workers ina colony minus the removed workers). Ths calculation showed that removal of ‘active workere should sighficantly decrease colony actviy lvel (Wilcoxon Signed-Rank Test, p = 0.021), removal of inactive’ workers snould decrease mean inactviy level (Wilcoxon Sighed- Rank Test, p= 0.031), and removal of random warkers should nat affect either colony actvly or nacbviy (Wilcoxon Signed-Rank Test,» = 0.50 and p = 1.00 respectively; S2 Table). Furermare, mean colony activity (te average ofall worker activity levels for hitps:iournals plos.org/plosonelartila?id=10.137/journal pone.0184074 ana2479122, 28:09 Who needs lazy workers? Inactive workers act as a ‘eserve’ labor force replacing active workers, but inactive workers are nat. that colony) was les than the mean activity lvel ofthe removed “active workers (Colon = 0.18, Removed ‘active workers = 0.42), {and mean colony inactviy was less than the mean inactivity level ofthe removed ‘inactive’ workers (Colony = 0.64, Removed ‘active workers = 0.75), Workers selected for the random removals treatment had comparable activity and inactivity levels #2 colony activity levels (Activiy: Colony = 0.17; Removed random workers = 0.20; nacity. Colony = 0.83; Removed fandom 87; 82 Table. When ‘active workers ware removed, nether average worker activity nor inactivity were significantly different from pre-removal actviylinactviy levels (Eig 1). This 's consistent with the dea thatthe colony rapidly compensates forthe removal of acive ‘workers. Mean worker time spent active and inactive pre-removal, 1 weak after removal, and 2 weeks after removal of (a) active workers (20% most macive in each calony) (b)Inachve workers (20% most inactive in each color), and () randomly Selected workers (20% random workers fram each colony). Boxplots show median (bar), quariles (box), and extremes (whiskers) for bos illustration (all 'gures). “Model: LMM, fixed: Worker inactvt/actvty = Ta, random: Colony-Contrast Significance delarmined using Tukey post hoc tests In colonies where ‘inactive workers were removed, there was a non-significant increase in mean worker activity after 1 week and a return to pre-arsoval actviy lvels after 2 weeks (o = 0.049, Tukey posthac shows no signifcant contrasts). Inactivity levels creased 1 wock ater removals and remained low at 2 weeks post-removal (lg 1B). hore wore no significant changes in worker activity or inactivity post-removal of random workers (Eig 12), ‘eee aye ft na slain mera pstenol compare loa oth met senna wera pecemel? In colonies where ‘actve’ workers were removed, the actvty levels ofthe most active workers post-removal (top 20" percentile 1 and 2 weoks after) were not signifcanly diferent from the actity levels ofthe most active workers pre-removal (Fig 2A Ie Furinermore, average time spent on specc tasks dd not cfr signtanty between pre-tpmoval and 2 weeks post removal (Ei 13). When ‘inactive workers were remaved, the most nacive workers pastemoval tap 20" percentie 1 and 2 weeks after) were ‘gniicanty less inacva 2 weeks post-removal than the most inactva workers pre-ramoval (Fig 2B—rghl) When random worker ‘wore removed, acivilylovels of tho most active workers significantly dacreased and remained lowored, while inactivity levels of the ‘mast inactive workers ware net signicanlly affected (Eig 2C). hitps:iournals plos.org/plosonelartila?id=10.137 journal pone.0184074 ana2414123, 28:09 Who needs lazy workers? Inactive workers act as a ‘eserve' labor fores replacing active workers, but inactive workers are nat. 42) (left) Meantime spent active by the most active workers (top 20th percentle)preremoval, 1 week after removal, and 2 Weeks after removal of active workers. (b) (right) Mean time spent inactive by the most active workers (top 20" percentile) pro-removal, week after removal. and 2 weeks ater removal of inactive workers.) (lt) Meantime spent active by tne most ‘actve and (ight) the most inactive workers (lop 20 percentile) pre-emaval, 1 week after removal, and 2 weeks afler removal of randomly selected workers. “Model: LMM, fixed: Worker inactvyfactivty ~ Tal, random: Colony-Contrast Significance determined using Tukey past hos tests. ‘nfosuidoiorg/10.137'/journal pone.0184074.9002 Fig3. Mean te spent on specie tasks by the most active worker (top 20 percentile prereroval and bythe mos active workers of hose remaining) 2 weeks aor removal of active workore “Model LMM, fixed: Worker time on task ~ Tia, random: Colony. ntosuidojorg/10.137V}oumalpone.0184074.9003 th moat ace nd ce waar paceman? For individual workers, pre-removal activity levels did not predict post-emoval (2 weoks) activity lovels for any removal treatment. However, pre-emoval nactty levels ware positively correlated to pastremoval inact levels when ‘inactive’ and random ‘workers were removed (Fig. 4). Ths indicates that workor nacivly may be more consistent than worker activity ar worker time Spent wandering (pernaps looking for tasks), which is consistent with previous work (50). This could result rom the inclusion of tasks with low repeatabilty such as grooming and trophallaxls in the ‘active tasks [8 ‘2 Time spent inactive preremoval predicts time spent inactive post-removal when inactive’ and randomly selected worker ate removed (b and ¢), But not when ‘active’ workers are removed (a). Time spent actve pre-removal dd not predic time Spent active post-emaval for any treatment (a, b and c). Colores points highlight different colonies. Marginal (variance explained by ied eects, e. by time spent actvetinactive pre and post removal) and Conditional total variance explained by fixed effects and random eects, ie. including betweer-colony diferences) R* a) Active Removal_~Actvy: Marginal Re = 10.007, Congitonal R® = 0.120; Inacvity: Marginal R® = 0.028, Congitonal R® = 0.231; 5) Inactve Removal—Activity: Marginal R?= 0.008, Corltonal 2 = 0.130; Inactviy: Marginal R= 0.084, Conditional R? = 0.337; ) Random Removal-— ‘Acti: Marginal R? = 0.018, Conditonal 8? = 0.658; Inactviy: Marginal R? = 0.048, Conditional R? = 0.496, ‘Model: LMM, fixed: Worker actvtylinactivty 2 weeks post ~ Worker activtyinactily pro, random: Colony. ‘ntosuidol.org/10.137'//oural pone.0184074.9004 hitps:iournals plos.org/plosonelartila?id=10.137 journal pone.0184074 ena24/323, 23:03 Who needs lazy workers? Inactive workers acl as a reserve" labor force replacing active workers, but inactive workers are not. The 20% most active workers befor ‘active! worker removal consislad mainly of workers frm the Nurse and Forager task groups, vwnile after removals the mast active workers were mainly composed of workers from the Inactve and Walker task groups (comember that task groups reflect the role of workers pre-removal: Fig A), While there ware Inactive workers te tap 20% Active workers postremova inal treatments, the presence of Inactive in tre 20% mst active was much greater inthe active ‘worker removals (Fig 5A—eft) than in random or inactive removals (Fig SB and §C—f). Furthermore, tha 20% most activa Workers after removal of random workers were significantly less acive than pre-removal (Eq 2G—ef). Therefore, the Inactive ‘workers present in the 20% most active workers afer rancom worker removal were overall less active than the Inactive workers that make up the 20% most active workers after active worker removal fl HE comm 2} let) Prior to ‘active’ worker removal, the 20% most actve workers were mainly ftom the Nurse, Walker. and Forager task ‘groups, but 2 weeks after removals the mast active workers were mainly from the Inactive ana Walker task groups. 6) (ight) Prior to inactive worker removal, the 20% most inactive workers were solely trom the Inactive task group, while 2 woeks after Femovals the most inactive workers were stl mainly from the inactive task group, ut there were also Nurses and Walkers (Fisher's exact test, p<0.0001), ¢) Alter removal of random worker, colonles had (let) more nactve workers and loss Taragers in the top 20% most active, and (ight) ess inacive workers and more of each other task group Inthe 20% most inacive workers. * Fishers exact tosis comparing frequencies of task groups pre- and 2 wooks post-removal hfesidolorg/10.137'/}ournal pone. 0184074 q005, In the ‘inactive worker treatment, the 20% most inactive workers (lop 20” percentile) pre-removal were solely from the Inactive task group (as determined by hierarchical clustering analysis). Post-removal, the most inactive workers were stil mainly from the Inactive task group, but also included some Nursos, Walkors and Foragors (Eig 8B). There was a small increaso in Nurses, Walkors ‘and Foragers inthe 20% most inacve workers post-removal i all moval reaimerts(Eg.5- rant). Mean worker inactivity of te Femaining 20% most inactive workers post inactive removal was signicanly lower than that ofthe removed inactive workors (Eig ‘22—igh). This suggest that, although the 20% most inactive post-removal included Nurses, Walkers, and Foragers, these lly didnot decroase thelr activity lvel to compensate for removed Inacivas Discussion In this study, we test whetner colonies of Temnothorax rugatulus ants maintain certain proportions of highly actve and highly inactive workers inthe colony. We show that colonies maintained pre-remaval activity levels afer active warkers were removed, espite calculations showing tat activity should have been significantly reduced had the colony not compensated forthe loss of active workers. We also show that's the previously inactive warkers who became active subsequent to the removal of ‘aca’ ‘workers. This constitutes evidence forthe hypothesis that inacve workers function as a reserve against worker loss ands relevant to the goneral question of task organization and now the colony respands fo changes in task demands, including by task switchos and switches to being active [55.77]. Conversely, colony actvly remained at higher levels, and inactviy decreased, after ‘Inactive ‘workers were removed. Thus, the colony dd not appear to compensate forthe removal of inactive workers. Mean worker activiy and inactivity levels were not affected by the removal of random workers Overall colony activity level was reestablished within 1 week of removals, and ‘active! workers were effectively replaced by workers that increased their activity to levels comparable to those of removed ‘active’ workers. Replacement occurred relivey rapidly (<1 week) and was maintained 2 weeks after te removal of ‘active workers. This suggests that there are mechanisms ensurng & Certain proportion of active workers within the colony. (Our data also showed thatthe 20% most active workers pre-removal were mainly composed of workers from the Nurse ang Forage’ lask groups, while 2 weeks past-removal, tre most achive workers were mainly workers wha had previously (pre-removal) been in he Inactive ‘ask and Walker task groups. This suggests that workers that were inacive pre-removal etfectvely replaced lost ‘active’ workers. Interestingly, if may be that Walkers also playa similar role. This suppor the hypothesis that inactive workers (and possioly ants that are wandering around with na clear task) are Teserves’ that serve as an auxiliary [ado force that can help the colony react quickly to workload increases. Altiough his the most commonly clad explanation for inact, evidence hitps:iournals plos.org/plosonelartila?id=10.137 journal pone.0184074 ana24/3723, 28:03 Who needs lazy workers? Inactive workers acl as a reserve’ labor force replacing active workers, but inactive workers are not. supporting it had been lacking. In fact, each previous study that expllcty sot out to test the reserve worker hypothesis by increasing ‘workload [23.30.47] or removing active workers (51,62 64] faled to suppor it, instead showing tna ether colonies do nat acjust to Changes in workload, or worxersather than inactivas increase ther activity to compensate forthe inereased workload ‘The only study that has offectvely increased inactive worker activity lavels did so by removing al but the mostinactive workers in colonies of the ant Myrmica Kofoku [8]. This study proposes that inactive workers are workers that have high response thresholds that only become active when all lower threshald workers are removed. Arecent madeling study also suggested that i such highly inactive, high-treshold workers exst, they may be pat of an adaptive strategy in social insects to mtigae the effecs of large scale disturbances where most ofthe workers in the colory may be lst (57), though itis unclear what these disturbances may be or how frequently they occur in natural populations. Nonetheless, the ant studied here, Tomnothorax rugatulus, may be the fist species discovered in which inactive workers directly respond o the loss of active workers by increasing their actly levels to levels comparable to those of the removed workers, effectively taking over as the most actve workers wih the colony. Ths indicates that (at least some) inacive workers are not incapable of working when needed. I also suggests that, inthe ant Temnotharex rugatulus, inactive workers may be part of an Adaptive strategy to rapialy allocate workers to novel tasks and changing workloads thereby contibuting to colony flexi and robustness. When the 20% mast inactive workers were removed, mean worker inactivity levels and the inactvity levels ofthe most inactive workers remained lowered both 1 and 2 weeks after removals. This suggests thal, conrary fo ‘active’ workers, inactive workers are ot replaced last. Thus, the replacement of active workers likely does nat result from random fluctuations in activity level for al ‘workers, in which case ail removed workers would be replaced without the need for specie mechanisms. Both pre and pos removal of ‘inactive’ worker, the 20% most inactive workers are largely composed of workers from the Inactive task group, Suggesting thal warkers are nat swiching fom olner task groups {0 replace the removed! inactive, workers amor iinay dt nrma ean tron \Whon randomly selected workers wore removed, there were na signfcant effects on magn worker actly or mean worker inact. A sight trend of decreased actvy and increased in inacivty occurred inthe 1" woek, Out ater 2 weeks, both mean ‘worker actly and inactivity appear to have regained levels comparable to pre-removal levels. Addtonaly, the 20% most actve ‘workers decreased their acy levels 1 week after removals and remained lower than pre-remaval levels even after 2 weeks, thaugh thre isa eight, non-signifeant nerease betweon 1 week and 2 weeks. Broadly. random worker removals have minimal effects on normal colony function 2 colonies maintain ther acvy levels pastremoval, trough there is cisruption inthe activity levels of the most active workers. ‘All romovals seem to have cause an increase in Inactive workers among the 20% most active workers, though the effect is greatest ‘when active workers were removed, folowed by random removals, and then inactive removals, Removals also lead to decreases In Inactvas inthe 20% most inactive past-removal, The proportion of Inactives in tne 20% most inactive was sigh salon he Inactive removal treatment than inthe acve or random removals whien should be expected when removed inactive workers are fot replaced. Thus, the disturbance produced by worker removals soem to cause a lovsleve reallocation of worker actly lvls, but the overall eflect of the treatments (Inactves effectively replacing lost active workers, andthe lack of replacement of lst Inactives) can stil be detected Group fxiblty allows collective units to quickly adjust to changing environments. Colony-level exit is broadly thought to be tained ether through highly flexible workers (Le. generalists) tat can switch between tasks when needed (but have high Cognitive and sensory costs) or by highly specialized workers (inflexible by definition) allocated to various tasks as needed (high bfcrency offset by cost of learning or morphological adaptations, e... large mandibies in soles), Here we show that colony flexibity can be attained by tascless workers that effectively function as tasicbutfers [though our rests suggest that there are mechanisms in place for maintaining some colony homeostasis in activity level, we do ‘ot know what these mechanisms are, Our results are consistent with task allocation mechanisms tat allow for exible reallocation ‘of workers to changing task demands such as response threshols [18.78], foraging for work [7], and interaction rate mediated task allocation [H0| Mechaniems such as age/temporal polyethism [8182) and task allocation based on body sie (aioethism) [83] ‘hich are functionally inflexible (reversions are possible In temporal polyethism but require longer timescale) are les tkelyt0 be allay. Rosorve workers (0. tasicloss workers thal sorve as poo of spare workers) havo only boon tested for ina haneful of spocios (money bees, bumble bees, Polistes wasps [23.30.47,5% 62)) that may rely more strongly an less flexible task alocaton strategies, And th's may have been the reason that nactve workers in Yiese species di not appear to quickly pck up needed tasks. As such, ‘we sil do not have a sense of how prevalent reserves’ may be across sacialinsecs, We can predic that colonies having evoWved Inunpregictable and high risk environments may have more seleclve pressure to mainlain a pool of warxers to compensate for ‘worker loss, Adtonally, we know that lager groups tend fo have less inactive workers [8d] which may indicate that larger groups have proportionally lass need for inactive reserves’ Of course, at this paint we can only speculate and much additional research is needed to gain a broader understanding of reserves’ as part ofa task allocation strategy. ‘The problem of adjusting supply to demand isa common one and the strategy of maintalning a reserve to deal with ths isnot unique fo socal insects [85]. For exarspe, supply chains maintain stockpiles of products (warehousing or slack) to avoid shortages as demand increases [38-88], employers often employ contingent workers ftom extemal labor supply agencies to deal wih changing demand (88), end computer systems perform better when they alow for a reserve of processing power (butering) [B2.20). However, te problem faced by all of these Syslems is how to opimally organize the supply of reserve workforce such as to minimize the costs of maintaining these reserves. Ths involves making predictions about the fur state of tne system (e.g. what future supply and demand wil be) and cosls associated with either over or underestimating supply and demand. hitps:iournas plos.org/plosonelartila?id=10.137 journal pone.0184074 7324/3123, 28:03 Who needs lazy workers? Inactive workers act as a reserve’ labor force replacing active workers, but inactive workers are not. ‘There does not appear to be a consistently optimal proportion of resources allocated to reserves across systems (@.. the proparton of exible workers in countries (temporary, standby, replacement, and other such workers) can account Tor 6.6%-10.7% Of their active labor fore [BB], while in social insects between 50-70% of workers are inactive at any time (84). This is not Surprising, because the optimal amount of reserves’ will e highly dependent onthe predictability and varibily ofthe environment ile, stable and precitable systems wil require less ofa butler than systems where demand fluctuates widely and unpredictably. Thus, human organizations may have less reserves because they face more predictable environments than social insects, of perhaps human organizations do nat sufficinty account for variation and should be employing addtional flexible workers fo ‘ptimally deal wth varying environments. Social insects face a range of predictable environmental fluctuations (¢.., regular cycles of brood production (94.92) as well as Unpredictable ones (@.., weather [93], food availabilty (94, predation (95), pathogens and parasites (96,971). However, te main problem of ants is competion (explotation, interference, and apparent) with other ants (generally [98-102] and specifically inthe ‘genus Temnothorax [103,104]. However, we actually krow very litle about Now much workload (demand) fuctuates in natural in Social insect colonies, nor how colonies adjust Leiravalable worklore (eupply) to these changing demands, For example, for most Social insects we know very litle about how food or ther resource avelabilty Auctuates in natural conditions, or how common, workers ae lst to causes other than old age, such as predation or natural events. We also know very litle about the costs of inaccurately predicting how ether supply or demand wil vary in the future. Here we show that that colonies of Temnothorax rugatulus ants possess mechanisms ensuring the presence, and ifnecessary relatively quick replacement, of highly active workers, But rot of macive workers, Altnough inactive warker may serve a function ‘thin th colony, suon as acing as food stores [SB], oF as a reserve labor force, our results Suggest that these may not be Continuously necessary tasks that require mechanisms to quickly reestablish thei occurrence when workers are lsL, On the other hang, i purses and foragers (which are the main components of the actives) are lost, the colony seems to have mechanisms in place to quickly replace them and ensure that these lasks are Keplup al comparable levels. This makes sense because without foraging and brood care, workers and brood won' be fed, which wil ikely incur a large fitness cost to the colony very quickly. Thus, it'seems that colonias do nol seek to maintain homeostasis of colony actly in goneral, but rather that worker replacement ‘depends on the immediate necessity of the task Alinough inactives may serve a function forthe colony, there appear to be no mechanisms ensuring that a set proportion ofthe Colony be decicated to this task. Instead, workers alocated to reserves’ may be determined by worker age and physiology corresponding to described tats of inactive workers (.e., young reproductives or young ‘tepletes’ who may not be wel suited to ‘work (58)}. Thus inactive likely result fram other age-related processes such as inexperience, physical wulnerabilly, degraded physiologies, decreased metabolic rates and immune function, etc. [8]. or worker tummover frequency being decoupled from the frequency of fluctuations in colony workload [17 In these cases, the mechanisms creaing inactive workers are acting on much longer timescales and s0 removed inacive workers may only be replaced when the colony produces more workers. We also showed for te ft time that inactive workers can act asa reserve labor fore, eflectively replacing lost active workers [Aknough this hypothesis is cften proposed to explain highly inactive workers in social insect colonies, previous studies that have sought to lest ithave rejected it This discrepancy suggests that inactvly likly has diferent causes and functions in ferent species of social insects Supporting information ‘insdoj.oro/10.137 Vfournalpone.0184074.3001 (00cx) ‘$2. Tble. Mean activiynactty love of colonies (clonyeval activity = pr colony mean of mean worker time spent on tasks) and removed workers {mean of mean worker me spent on tasks for removed workers) fo each removal etmert. bps ore/10.137tjoumal pone. 0184074 5002 (ocx) {top Distribution of mean activity and inactivity levels across colonies (mean of worker activity and inactivity levels for each colony fas a data point) and (bottom) dstabution of mean activity and nactvly loves across workers (mean of absorved activiy ang inactivity levels for each worker as a data port, excludes workers only observed once). Journal pone.0184074.3003 (00cx) |SZEia.Removalof ctv workers andinactiesignlcatly decreased mean worker activity and inactivity levels respectively (upper left anu ‘ight while removal of random workers shoul nt sgnifiarty affect mean worker activ or activity (bttor le and eight). Figures show comparisons of activity and inactivity levels of whole colones (i.e, mean actu levels ofall workers in a colony) pre= removal o te caleulatod mean colony activiy level of colonies without the workers thal would ler be removed (io. mean activity lovols ofall workers in colony minus the removed workers). Because of lost stored samples, not al ramoved workers wore identiad, therefore these analyses only include a subeat of all data, “Wilcoxon Signed-Rank Test: Colony activyrinactviy level ‘with removed workers vs, colony actviylinactiviy level without remaved workers, btipssJdl ora/10,137Ujoumnal pone. 0184074 5004 (D0cx) References iw Arce = Goole Sable " 2. Janson S. Emergence: The canned ves of as trans, lesa stare New York,NY: Sinan and Schuse, 2012 MiwArce + Goole Sable ° hitps:iournals plos.org/plosonelartila?id=10.137 journal pone.0184074 ana24/3723, 23:03 Who needs lazy workers? Inactive workers acl as a Yeserve' labor force replacing active workers, but inactive workers are not. hitps:iourals plos.org/plosonelarticia? Rubensain M Coma, Nagpal 8, Pograileset-tsarbly ina thaardscbot nim, Sone, 20448: 78-199, gg 25124495 (ANSEO) 2011 2nd intraoral Conereree a, 20. pp 731-704 Lay Ran Ho, Nakao, Supa UT, Mend Mal Sbucureanfrmaon of at anaperan tw. 1 Soc ace 20178 MwAae + PubadNO « Goad Scale Durem E, The dvsn of borin soy New Yank NY: Sin ana Senor: 1897, Becker, MirnyAN.T vsn of br cordnton cats, ae nonlege,0 ean, 8207: 1437-100, MwAdce + Canoe Sale Reinet. Aspacs of rgarzatonarganagencas ad embryogenesis, Plant Tsue Cl Gu, 1973; 38-885, Ontos a death clay of cose yt rt Oxon Unverty Prose 210 Avnblbeiboks aoe comoks? Gowrie + PubledNot + Geoat Schope ‘Ailes A Moreno A Muti agency an eganzatona ew Bil Pb, 20150: 35-387 iowAice + Goce Sota BorabeauE, Sccalinset colle 6 comloxacplne ystems Eeosjtans, 19083: 437-148, iowAMice » ‘Gonle Sal ViowAtice » PubledNotl = Cosa Scale Carson 0, Domnaus A. hen dong robin 6 Sematig. How tsk abacation ssegles compromise tevoenfxbity, fiery endnactve swAle + Goons Saale, MewAnile + BNWeINOS Canal Snot Sealey TD. Th wacom on hve te soci physiogy of one be sores [eet Mararé Universi Press 2008. val: FWiGhsOkdsio ONSATBAGEOUecASISn SPs ‘Rear MD, Danaldeor Manet, Domhaus A uti ine: Whan ohana be rages deci alop foragng on nla esas? Frnt El Ev Dre ©, Tetpy DR. Percept fhe pen nse ty foxages a henybee exon. im Bev. 2000 $9: 91-36 pri 10640870 Vinnie * PubidedNo! = Gavel Shale Pan, Fond MK. Boo phorrtona sina palle ‘epg mt anay bees Ap8 melee, Beh ol Soca. "9084: 103-196, Mowe + Goole Seba 10.137 journal. pone. 0184074 ona24/323, 28:03 Who needs lazy workers? Inactive workers acl as a reserve’ labor force replacing active workers, but inactive workers are not. hitps:iourals plos.org/plosonelarticia? sore, gel, Ceestman IM, Fahiach SE, Rotinson GE, Tirekeesing nthe honey be calony grtn of reagan hth and isan of ‘MewAtice + Gace Schl: Gordon OM. aly Rynsin Saal Aceves fe Harvester Ar. Pogenomymex dus Payee JElamal, 1983.90: 13-423. owAtice + Goole Stl Nau. structure and sence ofthe soil neta in en lewArce +» Gooae Sete Duane A, Weeing Fu, Pan | Ke L.A Evotonary Prepctve on Sal-Orgenzed Dison of Labor Soc nec, Anu Re Eel Eval Syst Jonson BR Realoaton a abr honybee clos sing hest en: he lative oles of ak wing athe ataton of reso labor. Senay MwAtice + Gaone Sota Cook oN, 468 0. Saal conten ituences he tan an Wrashol of armeroglsny Behavior n honeybees Ann Bahay 2015: $2328 lwAce »” Goole Sete Postel 4, Oereuboug J, Gos 5, Setarganzaon mechanisms nant sodas. Ta ecultmen: one discovered oo sources. om niviualiocoleztve bomen socanesc os Tes Werkatapeded by Icnuet M Postel, earlaus Denutouy. Sal rkner 1887 907, Avolote heise ea arlanlesathieeth deci USOT NTO SecharsR,Deneubous J Goss 8, Pasa JM. Calece ection mag tneuph ood eset. nadas Sosa 1900; 258-287 MiewAltie » ‘Good Schl “rave JF, Rabon SK. Tl rte communication sca nsec, Chel Ecology of eds 2, Sng, 8B6 pp. 241-286, Ava! iowAtice + Goole Soba: huang 2% Rabison GE Regulator ethane be sn flab by cley age erogrshy Behay Eee Soci 199639: 67-188 ViowAice = ‘Goole Sal Gordon OM. Te aac ofthe daly curd of he ners ant ony (Pogonamymex bubstie} Ann Bahay 1686: 1402-140, iwAice + Goooe Sota ‘son EO. Cate and ison ofebor nacre (Hymenoptera: Forme: As). Behav oa Sebi. 183.16: 47-54 iwAcle + Gaoae Sate ‘son EO he elaton Belvo case rate ad isn of borne a genus Pra Hymne: Formcdee), Betay Eco Sade 198616: edenai, Topo Sox egatn of ahaverldevelpmania ta at, Nvoresralbzalaeus (Maye J Carp Psychol. 198599: 316 Janson BR. Organzatan of workin he honeybee: comromic belwoendvision of hbour ard behav axl. Proc R Soc and 8 Bl Se ‘lowAice + EubedNOB + Cosel Scot Senne Hompal Wore casts ane aden demoaratyJ Evel Bol, 19235: 1412. ‘wap Cl, Thin WR. inet an nv ongevty shape frageralocaten nthe lari harvest ant Pogononymex bad Bahay lewalite » Game Scale, Kenic Ci, Tchinkl WR. Deropapy, dean, dst, andthe sensonl alcetion f bor nthe Fae harvester ant (Poganomyrex ba) ‘lowArie + Goole Soba lowe + Goole Schl ore 0. Honay 268 coatan cocks behav! conta rom naval wera fo whalecloy ym J nsct Pye 200147: 84-867 MewAtie + Goode Sete 10.137 journal. pone.0184074 ron24/3723, 23:03 Who needs lazy workers? Inactive workers act as a Yeserve' labor force replacing active workers, but inactive workers are not. hitps:iourals plos.org/plosonelarticia? tna J Robina N, Moore R Damau nvidia bumblebees Wy espade lo datarc. 8 al of the de Soca, 2012-58 313-21 Gasagar 8 Josh NY Socal ogaraton nh nan warp Rapala cyatorms Fannie: Vespéo),Z Fr Nene, 184i 8-22, MuwAle * Gonna Sale. aleolo, Sean 6, Bebavoiral eerces betwsen ale md for rpacemantreprosucives Kaltes fails ope, Kalter). iowAice + Goce Sctle Careona 0, Damhaus A Woks epeized” or natty: Behav constancy native warts nd rl tat lace, Bahay Eee MewAlle + Goons Stale. Horbrs JM, Socal Ormarizaton fn Lopotorax Ars Wehindnd Blwoen Specks Pats. ryche J Erol 18890" 96-286, (Cate The scl behav of Lepchoranatrdyee (hymenoptra, Foie: tne bxges adhe evlon of worker producto, Beha El iowa + Gove Shei Retane Cra X. Sct rganzaton of Calapan curser Ar Coles (Hymenontae, Formdbe) Ie, ang spec Comparisons Ethology Cab? Behar oye Hyenoptars:are-examnaon othe conop of cate. Ady Myrmoca. 808 737-3 Prooessng in Soc Insets, ruse Basa 1990. 9p. 230-250, ProNoman N Hub J Holey 44, rans, Dontaus A. How I acy estbuls amorg ara wtb in Tethorax ais? Behav Eel MwAlte + Goons Saale. Charboneau 0 Pf Nguyen I Sn MC, Krad K,Dorshas A. Who Are he. Ars? The Funcn of natty in Sock Insets and Poste ale of Conta native Ata Are Capua May Be Young anlr See, ay Comp Bl. 2017, MwAice + Goose Satake A. Bumbles responds teetois and body ez: ds wort dary nore oly parumancn Am Bahay, 20887: 97= freed MO, wan DB, QueralA Doman fr ak pariomance ane wcrc placerat undarater honeybee, Ap rier, ene. Jeet, iow Ae + Goode Schl ‘ole. Bahse Eel Sansbi 00781 763-792 Mow tte +” Googa Sila Bovey 80, MeLendon H, Nau 5, Hames S, Gordon O. How Flt acs tonsa atrencasn fren atv of haves ais Bahay MewAne + Gaonle Salar MiewAile + Goole Saale. Lich Le comparamant ronson don tava chee a Fur Lass gr.) Univer do Tar. 1978 lah, Hasagans ET mechan underhing he repuaon of wa-sd behaviors inthe manomaric an Myc kalo, JE, 209;%:8.- ViewAMte + Goode Schl 10.137 journal. pone. 0184074 ans24/3723, 28:03 Who needs lazy workers? Inactive workers acl as a reserve’ labor force replacing active workers, but inactive workers are not. hitps:iourals plos.org/plosonelarticia? ‘all: lyn oe anh oslo PRCA Toa {carson 0. His W, Goma A. Aree" ais ash? Repraducve compton may bas workrs wad soli soneg ts, Aim Bah. ienay Exo Scab, 29TH 2941-200 MswAtce + Goole Saale eB. Oominane ararchss in potas Se NY NY. 881212 88 MiewAude'» Gone Stale. ‘Domne, Buibwa Sl Weer connec evow of te design of wore ommncatonayoms and ts lacs on ask perrmancsin insect lncnard GB, Ofedge GM, Reyokt SE, Franks NR. Dito of bour sestonaltyn beat Lepore aban: woke corpence ad MewAae + Pubes» Gasol Shol Ccharbonnoau 0 His W, rma, Lyin MiewArice + oe ar clay teva sigs show hgh att” ha fal as wo a in th nots Sci. Franti $6, Gson Dl, Rabson PA, Potlnan J, Fah JS Pata nal arab or determin Sacet prinal components J Veg Sa swAle + Gone Sale: Lenoir Madan Jc Noe sur Lappland anaes sorespardarces a cvson aval are ourmis, Comptes Rencis Hed Seances iowAnice + Goole Sota Retane 1 Cat X. Bebavowal Vrably an Devlpart of Claghpis cor An Woks (Hymenoptera, Familae MewAle + Gonna Saale: tology, 109189: 275- Cab, Tonio f Seva Hsin age casts ofthe rhea denn. J Inset hay 19882 669-67 ewAlie'» Canoe Saale wiowAricle + Gos Franks NTs C Foti fr WorkHow Tasks Moca Werke. fri Be 190448 470-472 MiowAiite » Goode Stl Gordon DM. The xyanzaton of waka sociansect ables, Netw, 1966-80: 121-124 wre + Goone Sete Sesey TD. Adapvesgntoance eb ep pastem sacle in honeybee clones, eh Ecol Socobal, 10821: 287-203, ion Ace * ‘Googe Sette wanile + Goole Sole 10.137 journal. pone. 0184074 rana22a, 23:08 ‘Who needs ‘lazy’ workers? Inactive workers act as a reserve’ labor force replacing active workers, bu inactive workers are nat 187, CriapnarM, Peck, Bllng terse saps enn It. Lopet Manag, 2088-14 Poe ME Alp 0, Tan negra worse capac and vento ranagerant under nour spline Int Prod Ras. 20047: 4201-04 Comping, 1987 Precedings 1997. pp. MBS7.10MPDC 199782237" aX, inet, Lee J Yu. Fete cove apt trough ative bulletin, Pall and Diba Procesting Symposium, Prooscngs arin GP season! Me cyl oft Arpaia, soma hums (ymeneptr: Formica. suthem Caled Arn Ente Se A, ewAtle + Goole sober Pot 80 Tacikel WR Fe antpalyorphis the atgnaries of bos product, Baha Eel Sob “965,16 323-36, us 5. Cost the cays ant Tetras asin yrancplea Forde) of ovrwatang aboveground, Eur Emo 201511017 wewAnive * Gos Medineir AP Folk 5, Schet | Stanton anutance Inte a Tenethrat hylndel vende on group Sze bol sie and acces tole Py MewAlie + Gonna Stale: Nona PO LM. Morially Risks. Food Quay Tade-Os na Carmen Cureny: A Patch Proferonces eso, 190,71; 1888-1882, MewAlie + Gonos Sale. SchmiéHompalPrasosin soci noc Inara, Prncson Uist Press 198, Ava seni ia Gs PPB achi asl acaba Aho ag VhvazPoaa SEK Bots, Mortnernnaz R, Baris L Means A Higes M Noseme spp ecn ands negave els an honey bees (As lira Deans et iewAlte » Goode Sta elon M1, lps GO. Stone cropping by Conomyrma tcl Hymaropea: Fema: Ane tohninus ofneorrceconpotitor- Behav Eos) Human KG, Gordon OV. Esllatn an aterence competion elneen he nese Agent a, npn hue, ab nalve St spe, MewAdibe + Pubes + Goaole Scola lewAice » Caoile sale: wowace » Gor Fite, Herer JM Cobnysbucle og sleverating at IL Fequeny a sve a 32pm 309089 Lach LP Abb K. An acon. Onto Universiy Press 2010. gal 2 e hs pepo, Evin. 200% 85.516 MewAlie + Puede + Gondola “Tematras gas, Gabor Eel Scab 20°50. 1265-127, wae = Goole Sable: hitps:iourals plos.org/plosonelarticia? 10.137 journal. pone. 0184074 sana