Dynamics of nest occupation and homing of solitary bees in painted trap nests

1. The oil‐collecting bee Centris analis (Fabricius, 1804) is an important pollinator for the Neotropical region. The species can be attracted to nest in human‐made cavities. Such trap nests or insect hotels offer the opportunity to study the behaviour of populations in semifield conditions.


Introduction
Trap nests are a commonly used tool for the study of above-ground nesting bee and wasp communities and have provided means to study the optimisation of trap-nest sizes in pollination systems (Oliveira & Schlindwein, 2009), nesting dynamics (Moure-Oliveira et al., 2017), rate of nest colonization (Martínez-Núñez et al., 2019), and foraging ranges (Gathmann & Tscharntke, 2002). These behavioural aspects can help to understand mechanisms of nest occupation and homing behaviour of solitary bees.
In this study, we addressed questions regarding nesting dynamics of C. analis in a newly established trap nest aggregation. We tested if the rate of nest foundation was influenced by differentially painted nesting options and if painted trap nests helped females to locate their nests. We specifically tested the following hypothesis: i) the number of occupied nests differs between painted and unpainted blocks, ii) bees learn the location of nests with time and painting helps with the recognition process, and iii) the duration of foraging trips decreases with time.

Material and methods
The bee hotel (BH) was built in August 2016 on the campus of the Federal University of Grande Dourados, in Dourados, Mato Grosso do Sul, Brazil (22 ∘ 11 ′ 41 ′′ S 54 ∘ 55 ′ 50 ′′ W). It was composed of 95 wooden blocks, with a total of 1549 drilled cavities of 0.6 and 0.8 cm diameter and 12 cm length. Thirty-three blocks (35%) with 395 cavities were painted with different water based colours (green blocks = 7, white blocks n = 6, purple blocks n = 8, yellow blocks n = 8 and blue blocks n = 4, Metalatex Eco, Sherwin Williams, Brazil), 62 blocks (65%) with 1154 cavities were not painted (wood). Observations started in June 2017, in the beginning of the breeding season of C. analis, when females were starting to establish new nests.

Number of occupied nests differs between painted and unpainted blocks
We monitored the nesting dynamics of the aggregation weekly from June to November 2017, by counting the number of established nests in each block. To test if the number of occupied nests was different between painted (human perception) and unpainted blocks, we used a generalised linear model with the ratio of occupation in a block as response variable (using the cbind function) and colour as predictor variable. Blocks were randomly assigned to treatment. We took the number of cavities per block as proxy for cavity density into account by including it as offset variable into our model. The nest cavities in our study had two different diameters (termed 'small' 0.6 and 'big' 0.8 cm), we took the different sizes into account by including the proportion of small nesting cavities as random factor into our model (glmer(cbind(number of occupied nests, number of empty nests) ∼ colour + (1|proportion of small nesting cavities), offset = log (total number of nests), family = binomial)).

Bees learn the location of nests with time and painting helps with the recognition process
Additionally, we studied the learning curve of females to locate their own nest after returning from foraging trips. For this, we individually marked 25 foundress females that were provisioning nests (observed carrying nesting material/pollen to the nest) with non-toxic markers (Edding ® ). In the afternoon when foraging activity ceased, the nest entrances of marked bees were blocked with a 1.5 μl Eppendorf tube, that was perforated to allow air circulation. At 6 am the next day, before foraging activity started again, the tube was removed, and females were allowed to forage normally. Foraging activity started around 8 am.
For the first five foraging trips of the day, we measured the duration (in minutes) of the foraging trips and the 'homing behaviour' which is the time (in seconds) a female spent hovering in front of the BH when returning from a foraging trip before it located and entered its nest. We used a linear mixed model to test if the interaction between the number of return event and block colour had an effect on the time a bee spent hovering in front of the nest entrance (lmer: time hovering ∼return event+ painted vs. unpainted + (1|bee)).

Duration of foraging trips decreases with time
In a subset of 12 marked females, we tested whether the bees reduced the duration of their foraging trips during the first five foraging events of the day using a generalised linear mixed model with negative binomial distribution and foraging time (in minutes) as response variable, number of foraging event (1-5) as fixed factor, and bee ID as random factor.

Number of occupied nests differs between painted and unpainted blocks
Centris analis females built a total of 104 nests (ca. 7% of the total number of available cavities) during our observations. In absolute numbers, the majority of nests were built in non-painted blocks, (n = 25) followed by nests built in yellow (n = 21) and purple (n = 21), blue (n = 16), white (n = 12) and green (n = 9). Yet, when taking into account the proportion of available and occupied nests per colour, our first model showed that the rate of nest occupation differed between painted and unpainted blocks (χ 2 = 34.344, df = 5, P < 0.001). Bees nested significantly more often in painted blocks compared to the unpainted blocks (P < 0.05) and proportionally more in blue painted nests (Fig. 1).

Bees learn the location of nests with time and painting helps with the recognition process
Our model showed a significant effect of number of returns from foraging events on the time bees spent hovering in front of their nest (main effect foraging event: χ 2 = 30.653, df = 4, P < 0.01). Painting had no significant effect (main effect block colour: χ 2 = 0.133, df = 1, P = 0.714) on hovering time.
Our results indicate that females of the oil bee Centris analis improve the relocation of their nests with time, suggesting a learning process (Lehrer, 1993). Upon leaving the nest for the first time of the day, all studied females displayed a recognition flight. The bees were observed hovering in front of the nest, probably performing memory retention before disappearing from the BH. When returning from foraging, the females were hovering again in front of the BH before entering the nest but reduced this hovering time during the 5 first return events (χ 2 = 19.767, df = 4, P < 0.001, Fig. 2) of the day by on average 13 s between the first return (R1 average = 17.15 ± 11 s) and the fifth return (R5 average = 4.29 ± 11.69 s). The time that the females needed to find their nests was higher after the first foraging trip, which also was found to be the longest foraging trip of the day (see below). It is known that bees might at least partially forget their nest location overnight (Dukas & Real, 1991;Keasar et al., 1996) and need to re-learn on the following day, resulting in an increased homing time after the first foraging trip (Dukas & Real, 1991 et al., 1996). Furthermore, it is known that bees use visual (Brünnert et al., 1994) and chemical (Guédot et al., 2013) cues for nest recognition and can remember the position of their nests by spatial memory learning (Collett et al., 2013) which can optimise homing behaviour (Ostwald et al., 2019). Differentially painted nesting blocks might therefore facilitate nest site recognition in a similar way. Cognitive processes related to colour have been found to be an important mechanism for bees in different situations such as in bee-flower interactions (Lehrer, 1993;Dyer & Garcia, 2014). This might indicate that the bees choose blocks of distinctive colours as a salient cue, in order to recognise their nest more accurately, which likely helps to decrease the possibility of drifting behaviour; a common feature in gregarious nesting bee species (Field, 1992). The use of painted nests has been found to significantly decrease drifting behaviour in honey bees (Dynes et al., 2019). However, if this is also the case in C. analis remains to be tested.

Duration of foraging trips decreases with time
Our model showed that the number of the foraging trip had an effect on the duration of foraging time (χ 2 = 13.684, df = 4, P < 0.01). The first foraging trip of the day was the longest, and the duration of foraging trips decreased with subsequent foraging events. However, this effect was mainly driven by the first two foraging events of the day. The first foraging trip of the day might have included an initial recognition and/or defecation flight (Cane et al., 2017) and foraging subsequently was optimised along the day.
To our best knowledge, this study with Centris analis is the first showing that painted trap nests play a role on nesting occupation of solitary bees. Even though our study is limited to one population of C. analis, we believe that our findings open to a new and fruitful line of research on the behaviour and ecology of trap nesting solitary bees.