Introduction

Climate has an important influence on ecosystems, and the changing climate has resulted in increased incidences of extreme weather events, including heatwaves, droughts, and severe storms. Ample evidence indicates that these extreme weather events not only affect the composition of assemblages in different habitats, but also impact the social structure of many animal societies (Walther et al. 2002; Merilä 2012; Sueur et al. 2019). According to the IPCC (Intergovernmental Panel on Climate Change), climate change has already exacerbated extreme events in Australia, such as increased incidents of flooding, which can have devastating impacts on Australian communities and ecosystems (IPCC 2022). Many wildlife species are well adapted to respond to periodic natural disasters, including floods (Weiskopf et al. 2020). One behavioural response to these extreme events is for species to relocate to higher elevations, where conditions are more suitable for their survival (Ma et al. 2019). Multiple reports have highlighted recent insect population collapses and a global insect population decline linked to extreme weather events and a changing climate (Lister and Garcia 2018; Sánchez-Bayoa and Wyckhuys 2019; Green et al. 2021). In this context, it is of particular interest to ask how insects may respond to the varied and complex effects of contemporary climate change as they are the most diverse lineage of multicellular organisms on the planet. Insects, which include many social, nest-making clades such as the estimated 20 quadrillion ants on Earth (Schultheiss et al. 2022), are of fundamental importance to the functioning of terrestrial ecosystems.

Ants spend considerable amounts of time and resources building and maintaining their nests as these structures protect them and their developing offspring from environmental adversities (Detrain and Deneubourg 2006; Gordon 2007; Dussutour et al. 2008). In eusocial insects such as ants, eusocial wasps and bees, and termites, the nest not only provides shelter to the adults and storage space for resources such as food, but also serves as a platform for performing communal activities and offers defensive benefits (Wilson 1971). Adverse environmental circumstances can affect a colony’s nest, decreasing fitness and, at the extremes, may force the colony to relocate (Nathan et al. 2008). Nest relocation in response to adverse circumstances is an integral part of the ecology of ants (Adis and Junk 2002; Nielsen 2011).

Flooding can be a catastrophic event for a colony, resulting in the drowning of the colony members, destruction of brooding chambers and disruption of many of the colony’s functions, potentially leading to the colony’s collapse. In tropical rain forests, floodplains, and intertidal zones, one of the many factors that induce ground-nesting insects to relocate is flooding events (Forschler and Henderson 1995; Lude 1999; Osbrink et al. 2008). Adaptations that enable these organisms to survive nest flooding have direct implications for their continued propagation. When nest cavities are destroyed by floods, social insects may seek refuge in drier and safer environments (Adis and Junk 2002; Nielsen 2011). For example, colonies of the Argentine ant Linepithema humile Mayr, 1868, are seasonally polydomous, and changes in humidity drive nest movement (Heller and Gordon 2006). In the case of a severe flood, L. humile relocates faster than during a typical move. During a flood, this species also chooses a dry location more often than does the odorous house ant (Tapinoma sessile Say, 1836, of North America, which often splits its colonies into two when flooding is high. T. sessile colonies split because some workers move rapidly to a new location while others remain in the nest and move deeper into the lower chambers (Scholes and Suarez 2009). In many solitarily foraging ant species, colonies rarely move their nest, and nest sites may stay in one location for many years. Nest disturbances, however, may trigger relocation of the nest in red honey ant Melophorus bagoti Lubbock, 1883, (Schultheiss et al. 2010; Deeti and Cheng 2021a). In addition, laboratory experiments showed that damage to the nest structure may cause nest relocation (Franks et al. 2006).

In recent years, the nocturnal bull ant species Myrmecia midas has attracted increasing attention for its cognitive, learning and navigational abilities as this species conducts visual navigation during twilight foraging, when light levels are orders of magnitude lower than the day, making detection of visual cues difficult (Freas et al. 2017; Narendra and Ramirez-Esquivel 2017; Islam et al. 2020, 2021, 2022, 2023; Deeti et al. 2023, 2024). This species is endemic to the coastal regions of Australia. There are about 90 species of bull ants in Australia with diverse behaviours and life cycles (Reid et al. 2013). While its navigational mechanisms and foraging ecology have been the subject of a number of studies, its life history has rarely been documented (Freas et al. 2017). Relocation in this species makes an interesting phenomenon to examine as nocturnal, solitarily foraging ants have not received any attention when it comes to this topic. In Sydney, New South Wales, Australia, our study area, nest flooding is a rare threat for these bull ants. Rainfall, however, was unusually heavy from February 2022 to March 2022, measuring 44% above average, with some regions of eastern Australia recording the wettest first three months of the year ever recorded (http://www.bom.gov.au/). We surmised that this extreme wet spell would damage some of the colonies’ internal structure and induce colonies to initiate colony migrations to escape flooded nest sites. As a natural experiment, we thus characterized the frequency of nest relocation in response to heavy rain in M. midas and the nest characteristics of both the old nest and relocation sites associated with these movements.

Methods

Study species and location

The nocturnal bull ants Myrmecia midas, are widely distributed across the coastal areas of Australia (Clark 1951). They are distinguishable by their large size (up to 3.5 cm), long body and massive mandibles at the front of their head. Bull ants are gathered in smaller colonies compared to other ant species, with nests that typically number in the hundreds of individuals. At our study site, this species’ nests are usually found at the base of eucalyptus trees (< 0.50 m). Foraging activity occurs nightly year-round, beginning right after sunset in the evening twilight, with foragers returning to the nest before the morning twilight (Freas et al. 2017; Lionetti et al. 2023). In the current study, multiple nests were examined in an open parkland in the northern part of the Macquarie University campus, in Sydney, NSW Australia (33.4614 S, 151.0639 E; Fig. 1). The observations were conducted from January 2022 to July 2022. Currently, there are no ethical regulations in Australia regarding research on ants, and these non-manipulative observations are expected to cause no harm to the species.

Fig. 1
figure 1

A topographical map of Macquarie University with the selected field site outlined in blue. All the activities of nests were observed in the highlighted area (Image generated using the NSW mapping service Sixmaps.)

Full size image

Procedure

We recorded the location of 50 colonies present at the field site during December 2021. In order to understand the foraging ecology of M. midas in their wild terrain, nesting trees were marked over three consecutive days during the hot start of the year in January 2022. We confirmed the colony presence by observing the workers maintaining the nest or foraging. Upon confirming the activity of the nest, we marked the colony location by placing a push pin on the nest tree. In February 2022 we took photos of each colony and recorded the positional coordinates of the nest using the Google maps app with a mobile phone camera placed on a tripod from one meter in height. A 360-degree compass (SILVA) was also placed next to the entrance on a spirit level facing towards the nest tree while capturing the nest location to record the direction of the colony in relation to the nest tree.

To examine colony relocation or split, the known colony entrance locations were observed from January 2022 to May 2022. Usually, during the Australian summer months, bull ant colonies are highly active, with a high number of workers leaving each night to collect food for the colony, maintaining the colony entrance architecture, and, occasionally, excavating debris from the nest entrance. During the period from March to May 2022 (http://www.bom.gov.au/climate/averages/tables/cw_066156.shtml), however, extreme rain fell throughout the area. The state of New South Wales recorded the highest 3-month rainfall in Australian history (http://www.bom.gov.au/). To discover the effect of such environmental conditions on this surface-dwelling arthropod’s relocation frequency, we obtained weather information from the Australian Bureau of Meteorology (http://www.bom.gov.au/). Heavy rain fell from the March 5 to March 9; during this period, we monitored any unusual activities of colonies. Most of the colony relocation or split occurred on March 10 to March 15. If a nest initiated a relocation or split, we also recorded characteristics of the relocated nest site. Once all relocating nest members had moved to the new site, the distance between the old and new nests was measured using a measuring tape (SONTAX). We also measured both the old colony site’s and new colony site’s characteristic features: the distance between the nest entrance and the nest tree (Distance), the diameter of the nest entrance (Diameter), and the height of the nest entrance relative to the surrounding ground level (Elevation), based on the height of the entrance mound. If, in two rare instances, the nest did not relocate to a new tree but instead nested in an open area and no nest tree was available, the distance to the closest foraging tree was measured for the distance-to-nest-tree metric.

To investigate possible reasons for relocation, excavation was performed on five of the relocated colonies’ old entrances of the nest. To expose the nest, a pit measuring 50 cm in width and 50 cm in depth was dug at one metre from the centre of the entrance. The pit was slowly widened to expose the chambers of the nest horizontally, one at a time starting from the top. From each exposed chamber, we looked for any damage in the tunnel and any dead larvae or other matter of the colony. We recorded the measurements of the colony diameter, total number of tunnels and depths of tunnel from the surface. We report measurements in the form of mean ± standard error of the mean in cm.

Data analysis

We used generalized linear mixed models to address our two major questions: to what extent do nest variables predict nest movements, and to what extent do climatic variables predict nest movements. We created one generalized linear model to test the effect of nest variables (Elevation, Distance, Diameter) on nest movements by creating a binomial response variable comparing nests that did not move to nests that underwent relocation or splitting. We started with a model containing all three-way interactions between our predictive variables. We then removed one by one each non-significant interaction until we were left with a model that contained only individual predictive terms and their significant interactions. We compared each step of dropping terms using Akaike Information Criterion (AIC) to ascertain and report the model with the lowest AIC value. Where we report the significance values of dropped interactions, we report them from the model immediately before they were dropped. We report the significance values of all predictor terms from the final selected model. We performed Tukey post-hoc tests to evaluate whether means differed significantly between the conditions before vs. after relocation. We plot the model’s predicted means with standard error for the significant terms and interactions in each model. The nest features, nest distance to tree, elevation and nest entrance diameter (area), were used as predictive variables. To understand the impact of climatic variables, rainfall, temperature and solar exposure were used as predictive variables. We compared 14 generalized linear models to test the effect of these three climatic variables on nest movements, models differing in the delay between the set of climate variables and the day in question on which ant relocation behaviours were observed. The climatic variables data was collected from the Bureau of Meteorology (the BoM, http://www.bom.gov.au/) on a daily basis in the observation period. For each date, the total daily rainfall, average temperature and total daily solar exposure were extracted from the BoM. We used sets of 2-day, 3-day, 4-day and 5-day averages of each climatic variable in the modelling (Tables 1, 2, 3 and 4): a) average of the day on which the ants’ relocation and split behaviour was tabulated (day 0) and the preceding days (Tables 1, 2, 3 and 4). In each case, we started with a model containing all three-way interactions between our predictive climatic variables. And as with the nest variables, we then removed one by one each non-significant interaction until we were left with a model that contained only individual predictive terms and their significant interactions. We reported the model with the lowest AIC value. Data were characterized and visualized in R (version 4.2.1; R Core Team 2021) using the packages trajr (McLean and Skowron Volponi 2018) and Durga (Khan and McLean 2023).

Table 1 Set of two days average climatic variables
Full size table
Table 2 Set of three days average climatic variables
Full size table
Table 3 Set of four days average climatic variables
Full size table
Table 4 Set of five days average climatic variables
Full size table

Results

A total of 50 nests of nocturnal bull ants Myrmecia midas, were observed over a period of heavy rain. Initially, the colonies affected by the heavy rains showed unusual behaviour in which some of the colony work force stayed outside on the dry areas around the colony even in daylight. For temporary shelter, some of the affected colony’s work force climbed the nest tree and hung on the tree surface, while others were found hiding under the bark of the tree or living under nearby sticks on the ground around the nest (Fig. 2). This behaviour is highly atypical for this species as daytime activity around the nest is usually minimal while foraging is restricted to after sunset (Freas et al. 2017). After all these events, we found that 9 of the affected nests (18%) had moved to a different place and 4 colonies (8%) had split into 12. We observed differences in the distances between the split colonies, ranging from 1 m up to 11 m. These 12 split colonies’ initial nests were also abandoned.

Fig. 2
figure 2

The effect of heavy rains on Myrmecia midas colonies. Some of the colony work force stayed outside in the dry areas around the nest. For example, some work force stayed together on a tree surface by making a raft (a), hid under the bark (b), stayed around the nest or under small vegetation (c), or stayed under nearby sticks close to the entrance by forming a raft on it (d)

Full size image

We first compared the nest variables, nest distance to tree, elevation and nest entrance diameter (area), between the original nests of those colonies that had relocated or split and the nests of colonies that remained stationary (Fig. 3). The model cast the nest variables as predictors while Non-relocated vs. Abandoned was treated as the dependent variable. The generalized linear model found that overall, the nest variables together were significant predictors of whether a colony relocated (F4, 60 = 19.4, p < 0.005, AIC = 48.2). Compared to non-relocating colonies, the abandoned nests showed significantly lower nest elevation (F1, 64 = 49.7, p < 0.005) (Fig. 3a), longer distance from the nesting tree to the nest entrance (F1, 64 = 20.8, p < 0.005) (Fig. 3b), and larger nest-entrance diameter (F1, 64 = 15.8, p < 0.005) (Fig. 3c). In predicting nest relocation, the model also revealed significant two-way interactions between nest elevation and nest-entrance diameter (F1, 64 = 14.2, p < 0.005), and nest-entrance diameter and distance to the nest tree (F1, 64 = 6.2, p < 0.01). Finally, the model also showed a significant three-way interaction (F1, 64 = 8.9, p < 0.005). We then compared the nest variables, nest distance to tree, elevation and nest entrance diameter (area), between the old, abandoned nests of relocated colonies and the original nests of split colonies on the one hand, and the new nests of these colonies on the other (Fig. 3). The nest variables were treated as predictors of Abandoned (old) vs. Relocated (new), which was the dependent variable. The generalized linear model found that the nest variables together made significant predictors of old vs. new (F4, 33 = 26.4, p < 0.005, AIC = 15). Compared to the rain-damaged old nests, the entrances of new nests were at higher nest elevations (F1, 26 = 118, p < 0.005), shorter distances from the nesting tree to the nest entrance (F1, 26 = 22.6, p < 0.005), and possessed smaller nest entrance diameters (F1, 26 = 79.8, p < 0.005). The model also found a significant two-way interaction between nest elevation and nest-entrance diameter (F1, 26 = 11.7, p < 0.005).

Fig. 3
figure 3

The Myrmecia midas colonies’ distance from the entrance to the nest tree, the nest entrance diameter (area), the elevation of nest and the two-way interactions. The distance from the nest entrance to the nest tree (a), the nest entrance diameter (area) (b), the elevation of nest (c) of non-relocated, abandoned, and relocated nests are shown. Relocated nests include 12 nests that were split from 4 colonies that abandoned their old nest. The two-way interaction between nest elevation and nest-entrance diameter (d), between nest-entrance diameter and distance to the nest tree (e) and between nest elevation and nest-entrance diameter (f) on non-relocated, abandoned, and relocated nests. The half violin shows the distribution of bootstrapped differences, the solid dot shows mean, while the vertical bar shows 95% confidence interval of mean difference

Full size image

Generalized linear models were also run using the climatic variables, rainfall, temperature, and solar exposure, as predictors of whether a colony relocated or split, considered as a dependent measure (Fig. 4). Of the models with different lags, model i) (days –3, –4 and –5) performed the best. The model with the lowest AIC (F7, 23 = 3.6, p = 0.009; AIC = 16) showed that all variables had a significant effect on nest relocation (Rainfall: F7, 23 = 3. 27, p ≤ 0.005, Temperature: F7, 23 = 2.93, p ≤ 0.005, and Solar exposure: F7, 23 = 1.34, p ≤ 0.005). Nest relocation or splitting is associated with heavy rainfall, low solar exposure, and higher temperatures. The model also found significant two-way interactions between all pairs of climatic variables: Rainfall × Solar exposure (F7, 23 = 1.39, p ≤ 0.005), Rainfall × Temperature (F7, 23 = 1.08, p ≤ 0.005) and Solar exposure × Temperature (F7, 23 = 1.32, p ≤ 0.005).

Fig. 4
figure 4

The Macquarie University field site’s rainfall (mm) (a), temperature (°C) (b) and solar exposure (W/m2) (c) during the four-month (Feb – May) period. The continuous line shows average rainfall (mm), average temperature (°C) and total solar exposure (W/m2) of each day. In each plot, red dots indicate colony relocation events and black dots indicate nest-splitting events

Full size image

Examination of five old nests abandoned by relocated colonies showed evidence that the heavy rains damaged the internal nest architecture (Fig. 5). Excavations of abandoned nests revealed that the passage connecting nest channels and chambers to the entrance had collapsed. In the deeper channels, 50 cm below the ground, brood was stuck in the soil. The excavated nests measured in ground area 29. 3 ± 2.5 cm2, in depth below ground 31. 2 ± 3.6 cm, and in tunnel distance from the nest entrance 25.1 ± 1.7 cm (Fig. 5 and Table 5).

Fig. 5
figure 5

Nest excavation of the old nest of the relocated colony shows the breakage of the internal chambers of the nest (a) and cocoons stuck in the collapsed channels (b). The half violin plot shows nest distance of tunnels depth and area from the nest entrance (c). The half violin shows the distribution of bootstrapped differences, the solid dot shows mean, while the vertical bar shows 95% confidence interval of mean difference

Full size image
Table 5 Characteristics of nest excavation of the M. midas abandoned colonies
Full size table

The relocated colonies made new nests at distances of 1 m to 10.3 m from the old. colonies (Online resource 1: Fig. S1; 418.8 cm ± 318.7 cm M ± SD). Split colonies made new nests at distances of 1.7 to 8.5 m from the old existing colonies (Online resource 1: Fig. S1; 382.4 cm ± 224.3 cm M ± SD). The relocation distance did not differ significantly between the relocated and split colonies (F1, 18 = 0.09, p = 0.76).

Discussion

Nest relocations are thought to be uncommon in Myrmecia midas colonies as well as other solitarily foraging species that nest in semi-arid to arid soils in which structures are resistant to deterioration. We investigated the nesting dynamics of M. midas and its response to heavy rainfall. Our observations in this natural experiment showed that a period of heavy rainfall was associated with M. midas colonies relocating or splitting. Additionally, we found that multiple nest characteristics predicted whether a colony would move after the heavy rains: the distance between the nest tree and the nest entrance (being longer), the nest entrance’s diameter (being larger) and the elevation of the nest above ground level (being lower) all predicted relocation in this species at our field site. Furthermore, the relocated (new) nests the colonies created were closer to their nest trees and had higher-elevated nest entrances with smaller entrance diameters.

While the data of the natural experiment are correlational in nature, our interpretation is that heavy rainfall caused the relocating nests to move house. Even without a formal time-series analysis, Fig. 4 shows that all the relocations and nest-splitting took place toward the end of the heaviest period of rain. The lack of nest relocation and nest splitting at any other point during the observation period is telling. We did not examine all conceivable climatic and non-climatic variables but believe it unlikely that other causal factors were at play. Another piece of corroborative evidence of nest movement being due to rainfall is that excavations of a few abandoned nests showed, in all the examined nests, extensive damage caused by floods, a likely factor in the causal chain leading to nest relocation. It would have been more convincing if we had excavated nests that did not move for comparison, but even though regulations in New South Wales allowed this activity, we personally did not think that it was ethical to dig up functioning nests and possibly destroy them in the process. Considering the observed damage, it is hard to think of other factors than the excessive water from floods that would cause this extent of deterioration. Earthquakes and fires may be the only other natural events that can cause far more extensive damage, but no earthquakes or bushfires were recorded in the observation period at our field site. We thus interpret the causal sequence as leading from heavy rains to nest damage to the decision to move.

Even with the heavy rains, not every nest relocated; only a minority did so. This datum suggests other contributing factors at play. Our analysis of nest characteristics suggests a number of features that predispose a nest to flood damage. Of these, elevation of the nest mound above ground level and smaller size of the nest entrance both likely serve to reduce the amount of rainwater entering the nest. We think that proximity to a tree delivers a similar advantage, with the root system of the tree offering some guard against chamber collapse. The nest tree also provides some cover with its branches and a trunk structure that affords some protection for a nest piled up right next to it, with the root system near the trunk also increasing the elevation of the nest mound’s entrance, reducing the chance of rainwater pooling over the nest. In all these characteristics, we interpret their effects as arising from the reduction of water-based damage due to heavy rains that these features afford.

Eusocial insects do not relocate their colonies routinely because costs in time and energy are associated with nest relocation. Relocating a nest entails a labour-intensive operation requiring the moving of hundreds or thousands of adults as well as the carrying of immature young and any stored resources from one location to another. During these movements, colonies face the risk of fragmentation and loss of members and an increased risk of predation as well as increased risks to the queen (Snyder and Herbers 1991; Robinson 2014). Moreover, foraging is not only stopped during the move, but after the move, foragers have to learn new routes to find food resources, costing time and energy. Together, these drawbacks to nest relocation mean that it should only be undertaken when staying put at the current site costs even more. In our study, nest relocation appears to align with catastrophic failures in the nest chambers, with relocation ensuring survival of the colony by escaping these flooded sites during/after heavy rains. It is interesting to study further how relocation affects colony fitness in the long term.

Nest emigration is unusual for many solitarily foraging colonies with a large work force. Various exogenous and endogenous factors, however, may cause them to move nests. For example, the Australian desert ant Melophorus bagoti Lubbock, 1883 usually does not move nests, based on some two decades of various teams doing field work at Central Australian field sites. The field work over the years, however, produced two studies reporting nest relocation (Schultheiss et al. 2010; Deeti and Cheng 2021a, b). Some nest disturbance might have played a role in triggering the relocation in that species, although it should be noted that the field experimentation usually causes some disturbance in each season, with the setting-up of feeders, partitioning of the area around nests, and painting of foragers. In most cases, no experimental nests moved. In Myrmica rubra Linnaeus, 1758, drought and frost were implicated in colony relocation (Brian 1952). Increased shading has been shown to cause colony relocations of M. rubra and M. scabrinodis Nylander, 1846, (Brian 1956). The current species, M. midas is also mostly sessile, and nest relocation had not been observed over some 8 years of field work in the area. Outside of a small window of days after record rainfall, we did not find nest relocations in the current study either.

The division of four relocating colonies into twelve distinct ones remains largely unexplained. Two weeks after heavy rainfall, our observations indicated that all the colonies at their new sites were actively functioning. One possible explanation is that simultaneous pheromone signals prompted many workers to excavate new nests at multiple locations, resulting in multiple colonies being established. Additionally, the relocation period coincided with the time of reproductive activities, with giants (winged queens) emerging from the colony and building new nests. It is plausible that some of the split colonies might have been taken over by these giants. Further research is required to fully comprehend the splitting of colonies.

Our analysis of the characteristics of newly established nests found changes in this aspect of the ants’ extended phenotype. Nests were built with safer specifications, in the sense of having characteristics associated with a lower likelihood of nest relocation during the period of observation: having smaller nest entrances, being higher off the ground, and being closer to the nest tree. The cognitive bases for such changes in collective behaviours is an interesting topic for further study (Sasaki and Pratt 2018; Sasaki et al. 2019). In the harvester ant Formica podzolica Smith, 1900, excavations showed that nests with high elevations were larger and had more tunnels than nests built with lower elevations (Sankovitz and Purcell 2021). Studies on Diacamma indicum Fabricius, 1798, showed that during the monsoon, they increased the colony elevation, depositing more soil over the nest entrance as a ‘sandbagging’ adaptation to avoid nest flooding (Kolay and Annagiri 2015). In some species of ants, nest mounds are known to function as levees to prevent flood water from entering the nest chamber during shallow inundation (Lebrun et al. 2011). Moreover, some ants change the elevation across the year with its varying weather conditions, with monsoon having the most profound effect (Kolay and Annagiri 2015). To overcome the climatic challenges posed by monsoons, the Indian ant, D. indicum, makes shallow nests but modifies the entrance with decorations and soil mounds (Kolay and Annagiri 2015). Camponotus anderseni Smith, 1900, workers, on the other hand, prevent nest flooding by behavioural means: plugging the nest entrance with their heads (Adis and Junk 2002). Whatever the proximal explanations for such phenotypic changes may be, we surmise that the function of such behaviours is reducing the risk of nest damage from flooding.

Extreme weather and climate change have profound implications for species and ecosystem management. Some dramatic ecological responses to extreme events have been observed across individual, population and ecosystem scales. For example, cyclones can alter the onset of sexual maturity in turtles (Dodd and Dreslik 2008), and prolonged droughts have caused population collapse in koalas (Seabrook et al. 2011). Understanding how organisms have adapted to deal with extreme seasonal changes in their habitat, such as monsoons, can help us appreciate and evaluate how organisms might, or might not cope with the extreme weather events now attributed to climate change (IPCC 2022).

Conclusions

We characterized nest-relocation and colony-splitting behaviours of Myrmecia midas nests in response to periods of heavy rain in its habitat. Heavy rainfall destroyed nest chambers in nests that subsequently relocated over the few days following the period of heaviest rain. We found multiple characteristics of the nesting site to predict relocation, including (longer) distance to the nearest tree, (lower) elevation from ground level and the (larger) diameter of the nest. Moreover, excavations of abandoned colonies suggest that heavy rain led to collapse of nest chambers, likely triggering the relocation we observed in these nests.