Nesting and foraging behavior of Xylocopa valga in the Ejina Oasis, China

Xylocopa valga is extinct in Latvia and Lithuania and is critically endangered in Poland, and its distribution in the Ejina Oasis, China, is currently unknown. Studies on the biology of X. valga are scarce, and thus, conservation efforts for this species are currently limited. Here, we investigated the morphological characteristics, nest architecture, nest structure and food type of offspring in the nest cells of X. valga. This research was conducted in the Populus euphratica forest reserve in the Ejina Oasis, China, between July 2014 and June 2019. The primary investigation methods included visual inspection, photography, observation and measurements of nest anatomy, and examination of pollen stores by microscopy. We found that in the Ejina Oasis, China, X. valga builds its nests in the dead wood of P. euphratica. X. valga is univoltine. Its lifestyle varies from solitude to symbiosis. When many females nest near each other, several females may share a single nest entrance, based on which they build their own cells. The nests are branched. According to our results, there is a significant difference between the thickness of the inner cell partition and that of the outermost cell partition in the branched tunnel. In the P. euphratica forest area, the food for the progeny of X. valga is mainly composed of the pollen and nectar of Sophora alopecuroide and Populus euphratica. Therefore, X. valga and S. alopecuroides exhibit close ecological interactions in the P. euphratica forest ecosystem.


Introduction
Xylocopa, the members of which are known as large carpenter bees, is an important genus in the Apidae family [1][2][3][4][5][6]. In recent years, Xylocopa populations have been reduced or even become critically endangered and gone extinct because of habitat resource reductions and damage [7][8][9]. Insight into the nesting habits and foraging activities of these species is essential for habitat recovery.
There are 31 Xylocopa subgenera distributed worldwide, which include 470 species [1,10]. Numerous studies on the nesting biology of carpenter bees in the genus Xylocopa have been reported. The time of nesting and the selection of the nesting substrate are related to the species and distribution of carpenter bees [11]. For example, Xylocopa ordinaria [11], X. pubescens [12,13] and X. cearensis [14]  wide range of nesting substrates. In contrast, X. virginica, X. ciliata and X. artifex show some specificity in their selection of nesting substrate [15]. X. abbreviata nests only on Encholirium spectabile of the family Bromeliaceae [16]. Therefore, the availability of dead wood with a suitable density can determine the distribution and abundance of Xylocopa species [1,15]. Most nests of carpenter bees exhibit a clustered distribution [15,17], which is related to the homing behavior of carpenter bee species [18] or nesting substrate resources [11,12,15]. The nest serves as a shelter as well as a food storage location for bees, and it is a necessary structure for bees to raise their offspring [1,19]. With the exception of Proxylocopa species, which nest in the soil, Xylocopa species nest in dead wood, hollow stalks and bamboo [3,7,[20][21][22]. The preferred nesting location and substrate have an impact on the distribution and genetic structure of Xylocopa populations [11,23,24]. In addition, the diameter of the nesting substrate and nesting substrate resources define the nest structure of carpenter bees [25,26]. Xylocopa bees build nests in botanical materials, and their nests are classified into two structural types: the linear unbranched (straight chain) type and the linear branched type [1,19]. Linear unbranched cells are cells continuously distributed along the main tunnel. Generally, the cells extend in one direction or two directions from the nest entrance. When the position of the cell being built in one direction is restricted, carpenter bees will continue to extend the building of cells in the other direction. The carpenter bees that build their nests in hollow stalks and bamboo belong to this category [15,21,27]. In the linear branched type of nest construction, there are branches in the main tunnel, and cells are distributed continuously along the branched tunnels. The nests that most carpenter bees build in dead wood belong to this type [22,25,27].
Xylocopa valga is one of eight species in the Xylocopa subgenus of the Xylocopa genus [10,19]. Males and females of X. valga are similar in morphology (Fig 1). Females are black, and their wings display a purple luster. The mandible contains 2 teeth. The two spines on the outer apex of the hind tibia account for 2/3 of the tibial length, and verrucous protrusions are found on the hind tibia [19]. The male body is also black, and the wings are brownish-purple in color. There is no verrucous protrusion on the hind tibia of males. With the exception of a small amount of black hair on the anterior surface of the head, the body is covered with brownish-black hair [19]. X. valga is distributed in northern Europe, North Africa and Asia. In China, X. valga is the only species in the Xylocopa subgenus [19] and is mainly found in Xinjiang, Inner Mongolia, Gansu, Tibet and other northwestern regions [10,19]. According to the International Union for Conservation of Nature (IUCN), X. valga is extinct in Latvia and Lithuania, critically endangered in Poland, endangered in Germany and Slovenia, and vulnerable in Moldova, Switzerland and Ukraine [7]. Previously, X. valga was recorded as building its nests in dead wood or wood structures [8,19], and nesting locations of the species can be found in standing dead wood, such as wooden electric poles and indoor wooden beams [8]. When there is a lack of dead wood in its habitat, X. valga may choose to nest in thick soil with a southern exposure [8]. X. valga has been reported to forage on 95 plant species in 30 families [8]. Nevertheless, compared with those on other Xylocopa species, reports on X. valga are rare. In addition, to the best of our knowledge, no studies have reported on the nesting and foraging behaviors of X. valga distributed in China.
In this study, the nesting and foraging behaviors of X. valga in the Populus euphratica forest area in the Ejina Oasis, China, were analyzed by observing and dissecting the cells of X. valga and observing its food and pollen under a microscope. The results of this study lay a foundation for studying the role of X. valga as a pollinator and protecting this species in the P. euphratica forest ecosystem.

Research site
This research was conducted in the P. euphratica forest reserve of the Ejina Oasis, Inner Mongolia, China. Ejina Banner is under the jurisdiction of the Alxa League of the Inner Mongolia Autonomous Region in the center of the inland arid region of Asia, exhibits a strong continental climate and is located in an arid desert area. The geographical coordinates are 97˚10 0 23@-103˚7 0 15@E, 39˚52 0 20@-42˚47 0 20@N, and its total land area is 114,600 km 2 . Only 23,000 hm 2 of P. euphratica forest currently remains. The annual average temperature is 5.7-9.3˚C, and the annual average precipitation is 38.2 mm [28,29].
The field sampling and measurements conducted in this study were approved by the Environmental Protection Bureau of Ejina Banner (no. E20140305A).

Measurement of morphological characteristics
We captured homebound X. valga bees in the vicinity of their cells in the P. euphratica forest reserve in the Ejina Oasis, Inner Mongolia, from June 1 to June 10, 2019. The morphological characteristics of the bees, including their wingspan, body length, mesothorax width, head length, head width, and proboscis length, were measured using a Vernier caliper with an accuracy of 0.01 mm.

Observation of nesting locations and cell structure
From July 1 to July 20, 2014; July 21 to July 31, 2015; and May 22 to June 10, 2019, we carried out field exploration in the P. euphratica forest reserve and found X. valga bees flying around the dead wood of P. euphratica. We left marks at the sites of dead wood and then randomly selected dead wood lying on the ground to observe and record the nesting locations of X. valga within the dead wood and their nest entrances through visual inspection, photography, and video recordings.
We dissected 20, 3, and 20 nests in 2014, 2015 and 2019, respectively. Dissection was performed as follows. The surface bark was carefully cut open with a knife, and chips and dust were removed. A Vernier caliper was used to measure the length and width of the nest entrance [22]. The digging direction inside the nest was observed, and the corresponding tunnel was opened carefully. Once the nests were open, the following parameters were measured with a Vernier caliper: the total length of the tunnel, the number of branches in the tunnel, the length and cell number of each branched tunnel, the length and width of each cell, and the thickness of the partitions between cells. We simultaneously recorded the developmental stage of X. valga (from eggs to larvae, pupae and adults) within the cells [23] as well as the degree of wear on adult wings to determine the life history of X. valga [22,23,31].

Identification of pollen species in bee food
Between June 1 and June 10, 2019, flowering plants within the foraging zone of X. valga in the P. euphratica forest in the Ejina Oasis were collected (9 plants in total). Buds that were about to bloom were collected and placed in a clean sulfuric acid paper bag. Then, the specimens were brought back to the laboratory and placed in an incubator at 25˚C for 6 h. Pollen was picked up with a dissecting needle and transferred to a 1.5 ml centrifuge tube. Acetone solution was added, followed by shaking for even mixing. The pollen solution was then dropped onto a glass slide with a pipette. The morphology of the pollen of different plants was observed with an Olympus STM6 optical microscope (×40), and these morphological data were used as a reference for the identification of pollen on the basis of morphology in the food of X. valga bees [32,33].
During the same period, 18 nests with 35 cells were dissected, and bee food was collected from the cells (n = 35). The length and width of the bee food were measured, and the wet weight was determined. Each sample of bee food was evenly divided into three portions, and a 0.2 g specimen was weighed from each portion and dissolved in 5 ml acetone. This solution was shaken for even mixing and dropped onto a glass slide with a pipette. The morphology of pollen was observed using an Olympus STM6 optical microscope (the proportions of different types of pollen were determined under an ×10 visual field, and the plants were observed under an ×40 visual field). The observed morphologies were compared with the morphologies of the local plant pollen from the reference glass slides to determine the proportions of different pollen types in the bee food [33,34]. Three subsamples were taken from each sample of bee food, and three visual fields were observed for each subsample under the microscope.

Statistical analysis
Statistical analysis was conducted using Excel and SPSS 20.0. An independent t-test was used to compare the morphological indices of X. valga females and males and the thickness of the inner cell partitions and the outer cell partitions in different years. A paired-samples t-test was used to compare the thickness of the partition of the first cell with that of the outermost cell in the same nest. Pearson analyses were performed to determine the correlation between the length of the branched tunnels and the number of cells in the tunnels. p<0.05 was considered to indicate a statistically significant difference. Origin 2018 was used for plotting.

Nesting location and substrate of X. valga
X. valga builds its nests in the dead wood of P. euphratica in the P. euphratica forest reserve in Ejina Oasis, Inner Mongolia. The locations of the nests were variable. Nests of X. valga were found in standing dead trees, standing stumps and dead wood and branches lying on the ground (Fig 2). All nest entrances in standing dead trees and standing stumps had a southern exposure, and those in dead wood lying on the ground were located on the shaded side, where the direction was unknown (Fig 3A). The nests of X. valga exhibited an aggregated distribution in the wood of P. euphratica. In each piece of dead P. euphratica wood, dozens of nest entrances were generally observed (because the number of aggregated nests was large and the discarded old nests and new nests were interspersed, exact numbers were difficult to obtain). The distances between nests were short, with the closest pair located within 2 cm of each other (Fig 3B).

Development of X. valga in the nest
Between May 22 and May 31, 2019, females were observed to select locations for the building of nests at their nesting sites. Between June 1 and June 10, females collected pollen for fermentation to produce food and laid eggs. Among the 35 dissected cells that were observed, 65.71% contained eggs, 25.71% contained immature larvae, and 8.58% contained dead eggs or no eggs. Among the 24 cells dissected between July 1 and July 10, 2014, 95.83% contained larvae (39.13% of the larvae were parasitized), and 4.17% contained pupae. Among the 11 cells dissected from July 22 to July 31, 2015, 36.36% contained white pupae, 45.45% contained black pupae, and 18.18% contained newly emerged adults. Between August 22 and August 31, 2015, newly emerged adults were observed visiting flowers in the wild. Based on the developmental progress of the offspring as well as the outside activities of the descendant emerged adults, X. valga is univoltine, and offspring overwinter as adults.
Nest architecture and structure of X. valga X. valga burrows into the dead xylem layer of P. euphratica to excavate a nest entrance and then digs a tunnel inward to build a nest. The nest tunnel dug out by the first X. valga female in dead wood is the closest to the outer part of the xylem layer. With an increase in nest density, females gradually build their nests toward the interior of the xylem layer. It is difficult to determine the nesting history of a family in a dead wood segment with a high nest density. The nest entrances of X. valga were nearly circular, with an average vertical diameter of 11.42 ± 0.37 mm and an average horizontal diameter of 11.67 ± 0.36 mm ( Table 2). The nests belonged to the branched type, consisting of parallel branched nests (82.61% of the 23 nests dissected in 2014 and 2015) (Fig 4) and forked branched nests (17.93%). The number of  branches in the X. valga nests ranged from 1-5, with 3 branches being the most common, accounting for 45.83% of the surveyed samples (n = 24), followed by 1 branch (8.33%) and 5 branches (4.17%). The average length of the branch tunnels in a nest was 68.43 ± 4.37 mm ( Table 2).
The length of the tunnel excavated by the female generally ran parallel to the texture of the wood material. The newly emerged females used one branch tunnel of the maternal nest to continue to dig the nest tunnel inward (or forward) (Fig 4C). Several females used the same nest entrance. There was no communication between females, as they built their own cells separately. Alternatively, X. valga females could choose a new location to build a new nest (Figs 2  and 4). Among the 43 dissected nests, 5 were built by reusing the maternal nest, with a utilization rate of 11.63%.
The cell number per branched tunnel ranged from 1-5, with a mean of 2.52 ± 0.12. The most common cell number per branched tunnel was 2, accounting for 41.38% of cases, followed by 3 cells, accounting for 37.93% (Table 2). Pearson correlation analysis revealed a significant correlation between the length of the branched tunnel and the cell number per branched tunnel (r = 0.660, p = 0.001). The total number of cells per nest ranged from 4-10, with a mean of 7.35±0. 35.
After the excavation of a branch tunnel was completed, X. valga collected pollen and nectar to produce bee food, and the females laid eggs on the bee food. Next, wood chips were excavated near the nest entrance to form a cell partition in order to block off the cell. The cells of X. valga exhibited a long cylindrical shape. The average length of the cells was 17.29±0.25 mm, and their average diameter was 11.24 ± 0.20 mm (Table 2). A cell partition separated one cell from another. The inner side of the cell partition was coarse, and the outer side was smooth. The thickness of the inner cell partition of the branched tunnel was 2.32±0.09 mm on average, while that of the outermost cell partition was 6.61±0.52 mm ( Table 2; S2 File). There was a significant difference between the thickness of the first cell partition of the branch tunnel and that of the outermost cell partition (paired-samples t-test, df (1,19) = -9.125, p<0.001) ( Table 3, and Fig 5). There was a significant difference between the thickness of the outermost cell partition of the branch tunnel recorded in 2014 and that recorded in 2019 (independent-samples ttest, df (1,19) = -2.969, p<0.001), whereas no significant difference in the thickness of the first cell partition of the branch tunnel was observed between years (independent-samples t-test, df (1,18) = -2.753, p = 0.230) ( Table 3). There was a significant difference between the thickness of the first cell partition and that of the outermost cell partition in both 2014 and 2019 (independent-samples t-test, df (1,18) = -2.352, p = 0.030) ( Table 3). Double partitions of the outermost cell were also found in June 2019 (Fig 5D).

Bee food storage
X. valga females collected pollen and nectar, which they fermented to form bee food. In this study, we collected 35 pollen balls in total. The wet weight of the bee food ranged from 0.9926 g to 1.9950 g, with a mean of 1.1210±0.0385 g (n = 35) (S2 File). Robust measurements and observations were completed for only 12 balls due to damage during sample treatment. The bee food was nearly rectangular in shape. Its length ranged from 15.28-18.77 mm, with a   According to optical microscopy, the pollen of S. alopecuroides accounted for 84.63±1.21% of the total bee food, that of S. salsula accounted for 6.79±0.65%, and unidentified pollen types accounted for 8.58% (Fig 6).

Discussion
In this study, we investigated the nesting behavior of X. valga and its pollen collection for bee food fermentation in the P. euphratica forest reserve in the Ejina Oasis, China, and found for the first time that X. valga nests in dead P. euphratica in the wild. Additionally, we found that the population of X. valga is locally dense and shows the notable characteristic of nest aggregation. These characteristics of nesting are presumably associated with the softness and strong antirotting properties of the wood fibers of P. euphratica. According to a literature search, no reports on the nesting behavior of X. valga in China have been published. Nevertheless, our team has also investigated the nesting behavior of these bees in some regions of Xinjiang and Gansu. In Manas and Altay of Xinjiang and Gaolan County of Gansu, the nests of X. valga have mainly been found in wooden beams and livestock sheds, and its nest substrates include poplar, catalpa and willow wood. In addition, in Manas (86˚12.6044E, 43˚50.3961N; altitude, 1573 m) and Altay (86˚25.4975E, 48˚09.5713N; altitude, 592 m), we found that X. valga nested in soil. These observations, combined with those in the literature [8,35], suggest that X. valga shows high adaptability and flexibility in the selection of nesting locations. However, the differences in its nesting behavior in dead P. euphratica versus wooden buildings and structures and soil as well as the influence of these differences on X. valga population structure remain to be explored. Xylocopa species may nest in dead wood because they exhibit a strong preference for this type of nest substrate [15] or because of limited nest substrate resources [17]. Within the nest substrate, Xylocopa bees often form a high-density agglomeration of dozens or even hundreds of nests with a very short distance between nests [17]. Offspring normally expand the maternal nest to form multiple independent nests, and multiple female bees use the same nest entrance [15]. However, this high-density, aggregated lifestyle increases nest site competition and pressure related to nest entrance recognition, intranest behavior and natural enemy defense [17]. Under the conditions of high-density aggregation, X. varipuncta can rapidly recognize the nest entrance after foraging based on visual and olfactory cues [17]. How X. valga rapidly recognizes the nest entrance, avoids competition with individuals of the same species and avoids natural enemies needs to be investigated.
Bee cells are shelters that protect vulnerable offspring as well as food for the development of larval offspring [1]. X. valga constructs partitions to block its cells, which is similar to the behavior of other Xylocopa species [11,22,15,27,36]. However, unlike the pattern observed in other Xylocopa species [21,25], the partition of the outermost cell constructed by X. valga was significantly thicker than that of the inner cells. Furthermore, the outer partition of each main tunnel was significantly thicker than the innermost partition. Although the thickness of the first cell partition in the nests of X. valga in 2014 was comparable to that in 2019, the outermost cell in 2019 was significantly thicker than that in 2014. In addition, we found double partitions of the outermost cell in 2019. The underlying nesting mechanism that explains why X. valga significantly increases the partition of the outer cell remains unclear and should be explored in the future.
Pollen is the main ingredient in the food stored by the mother bee, which serves as the sole nutrition source for bee larvae [19,37]. Different carpenter bee species collect different types of pollen. There are also differences in the kind of pollen collected by the same kind of carpenter bee in different distribution areas. In Argentina, 18 kinds of pollen have been reported in X. augusti food [38]. Twenty-nine pollen types were identified in the food of X. nasalis, 13 of which were the main pollen sources [39]. X. valga uses a variety of plants as food sources [8]. This study was conducted in the Ejina Oasis of China, where the type of pollen found in larval food of X. valga was relatively homogenous. According to our results, the pollen of S. alopecuroides accounted for over 84% of the total bee food samples, which might be related to the species and distribution of nectariferous plants in the P. euphratica forest area. The influence of S. alopecuroides on the health of X. valga and the development of the offspring may be another interesting research direction in the future.
The P. euphratica forest in the Ejina Oasis is one of the main natural P. euphratica forests in the typical desert area of China and is the only forested area in western Inner Mongolia. The forest also constitutes an ecological barrier that helps maintain the Ejina Oasis and has considerable ecological and scientific significance [28]. S. alopecuroides, which is characterized by strong drought and alkali resistance and fast growth, is a pioneer plant used as a windbreak and for sand fixation in the P. euphratica forest area [30]. Its flowering period lasts from May to June, and its full-blossom period is from June 1 to June 20. Our study found X. valga nests only in dead P. euphratica and that S. alopecuroides in full bloom is the major food resource used by X. valga to nourish its progeny. Our results reveal that P. euphratica, S. alopecuroides and X. valga show a close ecological relationship, laying a foundation for further investigations of the interaction between species in the P. euphratica ecosystem.

Conclusions
This study is the first to report that X. valga nests in dead P. euphratica in the Ejina Oasis, China. X. valga is univoltine and lives alone in cells, with multiple females using the same nest entrance. The bees produce branching-type cells, in which the partition of the outermost cell is significantly thicker than that of the inner cells. In Ejina, the dominant nest substrate for X. valga is dead P. euphratica, and the dominant food source is S. alopecuroides. Additionally, X. valga is the dominant pollinator of S. alopecuroides. The results of this study may lay a foundation for future exploration of the relationships between the population structure of X. valga and its nesting and food resources. In addition, this study may provide useful data for future X. valga conservation efforts.