Introduction

Natural resources are a fundamental part of the economic development of a nation and represent a potential progress, if they are managed in a sustainable manner. The arid and semi-arid zones of Mexico provide resources which result in long-term economic benefits to rural communities (^{De Luna et al., 2013}; ^{Dinwiddie et al., 2013}; ^{Cruz et al., 2014}; ^{Lara et al., 2015}).

Insect harvesting is an ancient practice that is still conducted in certain regions of Mexico. Some insects are used in the gastronomy industry, and are currently considered as gourmet food (^{Ramos et al., 2006}; ^{Miranda et al., 2011}). Edible insects -like escamoles (Liometopum apiculatum Mayr), white agave worms (Acentrocneme hesperiaris W.) and red agave worms (Comadia redtenbacheri Hamm)- generate continuous employment for rural communities of the arid ecosystems of Zacatecas, Mexico. In the municipalities of Villa González Ortega, and in Pinos, Zacatecas, the harvest of edible insects can be a better alternative income than cattle raising and rainfed agriculture. It also offers employment options to the inhabitants, when no other productive options are available (^{De Luna et al., 2013}). The escamolera ant (L. apiculatum) occurs in 15 Mexican states, including: Estado de Mexico, Hidalgo, San Luis Potosí, Tlaxcala, and Zacatecas (^{Del Toro et al., 2009}); its distribution and abundance are limited by temperature and relative humidity (^{Cerdá, 1998}; ^{Cruz et al., 2014}).

The colonies of the escamolera ant commonly have one to six foraging trails; their length varie according to the habitat conditions and none of them crosses each other (^{Ramos et al., 1988}; ^{Lara et al., 2015}). Foraging is carried out between 7:00 h and 19:00 h., from March to September, in up to 580 m² per colony (^{Ramos and Levieux, 1992}). In this activity participates only 10 percent of the colony approximately (^{Dornhaus and Powell, 2010}).

The larvae of L. apiculatum Mayr are considered food in the states of Chihuahua, Durango, Michoacán, Colima, Hidalgo, Mexico, and Puebla, as well as in Mexico City (^{Ramos et al., 1988}; ^{Cruz et al., 2014}; ^{Lara et al., 2015}). In the San Luis Potosí-Zacatecas High Plateau, L. apiculatum represents an important economic resource during the drier months (March and April), when the reproductive breed is harvested in its larval stage (escamoles). The price of escamoles in the region varies from $150.00 MXN ($7.5 USD) to $550.00 MXN ($27.5 USD), with an average price of $250.00 MXN ($12.5 USD) per kilogram (^{De Luna et al., 2013}; ^{Dinwiddie et al., 2013}; ^{Lara et al., 2015}). In the Teotihuacán zone, Estado de Mexico, escamoles are sold per liter (approximately 800 g), at $750.00 MXN ($37.5 USD), (^{Miranda et al., 2011}).

Frequent land use changes and recurrent droughts have caused the loss and fragmentation of the habitat of this species. There is also mismanagement and lack of legislation regarding the escamolera ant (^{Tarango, 2012}). In the semi-arid ecosystems of central Mexico, L. apiculatum has been overexploited. As a result, their colonies have been destroyed and their population have declined (^{Ramos et al., 2006}; ^{Tarango, 2012}; ^{Dinwiddie et al., 2013}).

Despite the economic importance of insects in desert ecosystems (^{Hithford et al., 2008}), and the extensive distribution of L. apiculatum and the economic importance of its larvae, there is not enough research on this species. Therefore, there is a need to study its habitat (^{Rojas, 2001}; ^{Tarango, 2012}; ^{Dinwiddie et al., 2013}; ^{Cruz et al., 2014}). The information from this study is important for the sustainable management of the colonies of the escamolera ant and their habitat, in a frequent land use change context, which has caused the opening of the ant habitat for agricultural activities.

The objectives of this study were to determine the relationship between the distance and the number of trails with the foraging substrate of L. apiculatum, and to identify the forage and nesting substrates, and their foraging effort. The hypotheses were: 1) there is not direct relationship between the distance and the number of trails with the forage substrate of the escamolera ant; and 2) the use of substrates is homogeneous.

Materials and Methods

The study was carried out from June to August 2014, at Villa Gonzalez Ortega municipality, Zacatecas, in the southern region of the Chihuahuan desert (^{Giménez and González, 2011}), located at 22° 25’ and 22° 40’ N, 101° 48’ and 102° 06’ W, at a height of 2000 to 2400 m. The territory has an area of 411.8 km² and includes 11 894 inhabitants (^{author unknown, 2010}; ^{INEGI, 2010}). The climate of Villa Gonzalez Ortega is temperate and semi-dry (BS₀kw’) and temperate and dry (BS₁kw), with temperatures ranging from 14 to 18 °C (^{García, 2004}). Rainfall is higher in August (300-500 mm per year). The most abundant vegetation consists of microphyllus scrublands and grasslands (^{Rzedowski, 1978}).

Field visits were conducted every day, from 7:00 h to 14:00 h., and samples were taken from colonies of the escamolera ant and from its foraging trails. With the support of an escamol harvester guide, we were able to locate the nest through the identification of the foraging trails of the colony and their intersection, where the ant nest is commonly located (^{Miranda et al., 2011}). The nest was verified by introducing a steel rod into the nest, which was then impregnated with the characteristic scent (similar to fermented butter) of an ant nest. The vegetable substrateused by L. apiculatum to build its nest was recorded and the foraging trails of the colonies were counted. Starting from the first colony, the rest (n=31) were selected taking into consideration their proximity to the first nest, but without considering their cardinal direction, and their foraging trails (n=116) were recorded. The number of trails indicates the number of plants that the ant forages. Once the forage substrate associated to each trail was identified, a circular plot (20 m wide) was established around it, and the arboreal plants and shrubs were counted within it (^{Schreuder et al., 1993}). The nests, the foraging trails, and plots covered 73.82 ha of the sampling area. This represented 27.58 % of the area under study in Villa Gonzalez Ortega municipality, Zacatecas.

The relative frequency of plant species -palm tree (Yucca spp.), prickly pear cactus (Opuntia rastrera), agave (Agave salmiana), catclaw mimosa (Mimosa Biuncifera), silver prairie clover (Dalea bicolor), creosote bush (Larrea tridentata), mesquite (Prosopis spp.), huizache (Acacia farnesiana)- was calculated using the following equivalence:

The foraging preferences were identified using the following formula:

The nesting preferences were identified according to:

The frequencies of observation were calculated using Microsoft Excel® (^{Microsoft Office, 2013}).

The linear distance (m) of trails from the nest to the forage substrate was determined with a laser rangefinder (Bushnell, Pro 1600 Slope edition Kansas, USA). A rope and a measuring tape (20 m) were used to measure the actual curvilinear distance that L. apiculatum travelled from its nests to the substrates. Finally, the level of association between these distances was identified using a linear regression analysis (Microsoft Excel®; ^{Microsoft Office, 2013}).

Descriptive analyses (descriptive statistics tables and frequency tables) were used to determine the relationship between distance and number of trails with forage substrates and identification of: 1) forage substrates in the sampling plots; 2) foraging effort (curvilinear foraging distance/linear distance); 3) number of trails per colony; 4) linear distance from the nest to the forage substrate; and 5) curvilinear foraging distance. These analyses were carried out using the Microsoft Excel® software (^{Microsoft Office, 2013}).

The probability of occurrence of arboreal plants and shrubs within the plots was determined using a simple regression analysis, taking into consideration the distances (linear distance from the colony to the forage substrate and curvilinear distance to the forage substrate) as dependent variables and the abundance of arboreal plants and shrubs as an independent variable; x_i was the number of trails per colony of escamolera ant. In this analysis, trail frequency was categorized as follows: few trails (two), regular number of trails (three and four) and many trails (five and six). The analysis was carried out using the GLM procedure of the R software, version 3.2.0 (^{Maindonald, 2004}; ^{R core team, 2013}).

The probability of occurrence of number of trails in ant colonies was determined using a simple linear regression analysis (^{Truett et al., 1967}). The dependent variable was the abundance of each arboreal and shrub plant species; while the independent variable was the number of trails per colony. The model structure was adjusted with the stepwise proceeding, minimum ^{Akaike classification criterion (AIC; Akaike, 1969)} with statistically significant coefficients (p≤0.05), using the GLM procedure of the R software, version 3.2.0 (^{Maindonald, 2004}; ^{R core team, 2013}).

The relationship between the number of trails of each colony and the presence of arboreal plants and shrubs was determined using a principal components analysis (ACP; ^{Hotelling, 1993}) using the GLM procedure of the R software, version 3.2.0 (^{Maindonald, 2004}; ^{R core team, 2013}).

The association of the independent variables (forage substrates: palm tree, agave, and prickly pear cactus) with dependent variables (presence of L. apiculatum) was identified through Poisson regression (p<0.05) (^{McCullagh and Nelder, 1989}; ^{González, 2003}), using the GLM procedure of the R software, version 3.2.0 (^{R core team, 2013}).

The differences on the use of forage substrates by the escamolera ant, or the frequency of its use, was determined using the nonparametric Kruskal-Wallis test (^{Zar, 1999}). The information of the presence of the ant per substrate was organized in three categories: substrate 1 (palm tree), substrate 2 (prickly pear cactus), and substrate 3 (agave), using the JMP IN software, version 12.1.0 (^{Academic SAS Institute Inc., 2015}).

The graphic associations of the presence of the escamolera ant -along with the most used forage substrates- were generated through a simple correspondence analysis (p≤0.05) (ACS; ^{Cornejo, 1988}), using the XLSTAT software, version 2015.1.01 (^{XLSTAT, 2015}). To do that, the information on the presence of L. apiculatum per substrate was categorized as: low presence (prickly pear cactus forage substrate), medium presence (maguey forage substrate) and high presence (palm tree forage substrate).

Results and Discussion

In this study, samples were taken from 31 nests of the escamolera ant (Figure 1) in colonies with a total of 116 foraging trails (Ẋ=3.7 trails per colony). The presence and foraging activities of the escamolera ant depend on the weather (temperature and relative humidity), vegetation type, soil and its cover, anthropogenic disturbances and on the level of association of the ant with some plants. In the case of the cresote bush (L. tridentata), there is a low association level (^{Cruz et al., 2014}; ^{Lara et al., 2015}), although it is very common in the habitat of the escamolera ant in central Mexico.

Figure 1 Study area and location of nests of the escamolera ant (Liometopum apiculatum Mayr) under study in Villa Gonzalez Ortega, Zacatecas, Mexico. 

The average curvilinear and linear foraging distance travelled by L. apiculatum was 38.6 m (+/-17.18 m) and 24.3 m (+/-11.36 m), respectively. The average forage effort was 1.6 (+/-0.32). The most frequent plants in the 116 circular plots were O. rastrera, D. bicolor, M. biuncifera, and A. salmiana (Table 1).

Table 1 Plant density and frequency of observation in 116 plots evaluated in Villa Gonzalez Ortega, Zacatecas, Mexico. 

The number of plants with presence of foraging activity by L. apiculatum were Yucca spp. with 20.3 plants per hectare (63.8 %), A. salmiana with 6.9 plants (21.6 %), and O. rastrera with 4.7 plants per hectare (14.7 %). Meanwhile, the curvilinear foraging distance increased with the linear distance from the nest to the forage substrate (R²=0.80; Figure 2).

Figure 2 Relationship between linear and foraging distance in areas of foraging activity of Liometopum apiculatum Mayr in Villa Gonzalez Ortega, Zacatecas, Mexico. 

In very hot and dry environments -such as the arid and semi-arid zones of Mexico-, environmental and soil surface temperatures influence the foraging activity and the ant foraging distances. High temperatures can dehydrate and kill ants and even lead to the disappearance of the colony (^{Cerdá, 1988}; ^{Lara et al., 2015}). L. apiculatum workers are active 24 hours a day, but they seek food during daytime (^{Ramos et al., 1988}). Additionally, ants establish underground foraging paths, under dead leaves and herbs that cover the soil surface; meanwhile, in rocky terrains, they move through the cracks (^{Lara et al., 2015}). In our study, the distance to forage on Yucca spp. was the longest (Table 2). This indicates that the paths and distances that ants travel depend on the forage substrate (Fourcassie and Traniello, 1993).

Table 2 Relationship among forage substrate, distances, and foraging efforts by Liometopum apiculatum Mayr, in Villa Gonzalez Ortega, Zacatecas, Mexico. 

Foraging distance decreased inversely to the number of trails (Table 3). ^{Miranda et al. (2011)} reported a maximum of four foraging trails and ^{Ramos et al. (1988)} recorded three to five trails whose width and length were directly related to the age of the colony of L. apiculatum.

Table 3 Number and distance of trails per colony of L. apiculatum Mayr, in Villa Gonzalez Ortega, Zacatecas, Mexico. 

Liometopum apiculatum prefers to nest in A. salmiana and Yucca spp. (Figure 3). ^{Cruz et al. (2014)} also recorded that A. salmiana was a main nesting substrate. Likewise, ^{Lara et al. (2015)} pointed out that L. apiculatum usually builds its nests under A. salmiana, palm tree (Yucca filifera), red biznaga (Cylindropuntia tunicata), and thistle chollas (C. tunicata), in the San Luis Potosí-Zacatecas High Plateau.

Liometopum apiculatum uses other substrates for nesting, such as shrublands, grasslands, dead trunks, rocks, and stems of decomposing Yucca spp. (^{Miller, 2007}; ^{Esparza et al., 2008}). In our study, colonies with three to five trails were more frequent for foraging activities (Table 4).

Figure 3 Frequency of observation of nests of the escamolera ant (Liometopum apiculatum Mayr) in nesting substrates, in Villa Gonzalez Ortega, Zacatecas, Mexico. 

Table 4 Results of the logistic regression analysis for the number of trails of colonies of Liometopum apiculatum Mayr in Villa Gonzalez Ortega, Zacatecas, México. 

Foraging distance was the variable that most accurately predicted the number of trails per colony (AIC=147.05; p=0.0644). In other studies, L. apiculatum searched for food in areas ranging from 361.5 to 580 m² and that the longest distance that ants travelled to find food had approximately a radius of 100 m (^{Mackay and Mackay, 2002}; ^{Lara et al., 2015}). The first three PCA axes accounted for 98.08 % of the variance (Table 5) and identified the foraging distance, the linear distance, and the forage substrates as the most important variables that indicate the presence of the escamolera ant in the study area.

Liometopum apiculatum is omnivorous, although it prefers liquid food, such as extrafloral nectars from plants and honeydew from some insects (^{Miller, 2007}). Poisson regression analysis identified (AIC=3191.4) a significative association of the ant with palm trees, agaves and prickly pear cacti (Table 6). Therefore, the spatial distribution of L. apiculatum was correlated with the location of bushes and trees infested by scale insects of the Order hemiptera (^{Ramos and Levieux, 1992}), since the escamolera ant feeds via trophobiosis from the honeydew of these insects, as well as from insect pupae, crustaceans,annelids, molluscs, dead vertebrates, animal excreta, and nectar from Opuntia spp. flowers (^{Velasco et al., 2007}).

Table 5 Typical values of the main components (PCA) with eleven variables related to the sites of the escamolera ants (Liometopum apiculatum Mayr), in Villa Gonzalez Ortega, Zacatecas, Mexico. 

Table 6 Poisson regression results for the presence of Liometopum apiculatum Mayr per forage substrate, in Villa Gonzalez Ortega, Zacatecas, Mexico. 

Some ant species have selective habits during their foraging process (^{Cortes and León, 2003}). In Mexico, L. apiculatum follows this pattern and associates with xeric shrubland habitat (^{Esparza et al., 2008}; ^{Cruz et al., 2014}). The association observed between ants and forage substrates -palm trees, agaves, and prickly pear cacti- is connected with food, but also with protection against predators and high temperatures (^{Miller, 2007}). A mutualistic relationship occurs between the escamolera ant and O. imbricata, both species obtain a benefit, the former obtains food from the latter (inflorescence nectar) and the latter is protected from herbivores and seed predators, because the former is aggressive and emits an unpleasant odor (^{Mackay and Mackay, 2002}).

Forage substrates (palm tree and agave) nearby the colonies were infested with scale insects (Hemiptera). These insects insert their piercing-sucking mouthparts into the plants (Yucca spp., A. salimana, and Opuntia spp.); the escamolera ant touches this structure with their antennas, stimulating them to eject their honeydew (^{Gullan and Kosztarab, 1997}; ^{Delfino and Buffa, 2000}; ^{Lara et al., 2015}), of which they feed. On this matter, ^{Cruz et al. (2014)} and ^{Lara et al. (2015)} reported the infestation of A. salmiana with the pseudococcidae Dysmicoccus brevipes and Coccoidea fallen and the relationship of ants with the prickly pear cacti, which likely allows them to use its flower compounds as food and the plant as thermal protection. In addition to palm trees, agaves, and prickly pear cacti for foraging, other bushy plants have a strategic ecological arrangement that protects ants during their foraging activity (^{Ramos and Levieux, 1992}). Vegetation structure and soil cover have an effect on the requirements of the ant, areas with more vegetation cover offer greater availability of food and create favorable conditions for the development of humus over the soil, which, in turn, generates a microclimate that protects the ants (^{Ramos et al., 1988}; ^{Lara et al., 2015}).

Simple correspondence analysis showed an association between foraged substrates and L. apiculatum; additionally, this analysis showed the presence of ants in three sets (Figure 4). However, the Kruskal-Wallis analysis (H=2.00; p=0.3679) proved that there were no significant differences with their presence in the three forage substrates.

Figure 4 2D simple correspondence representation between the presence of the escamolera ant (Liometopum apiculatum Mayr) and forage substrates, in Villa Gonzalez Ortega, Zacatecas, México. 

Liometopum apiculatum prefers to carry out its foraging activities on Yucca spp., A. salmiana, and then on O. rastrera. Ants traveled greater distances and carried out greater foraging effort (Ẋ=1.6) to forage on Yucca spp. However, when colony trails increased, foraging effort diminished. Although the density of palm trees in central Mexico is lower than the density of prickly pear cacti and agaves, their volume is greater and offers foraging conditions all year long.

The settlement, subsistence, and persistence of colonies of the escamolera ant depend on the habitat (^{Lara et al., 2015}) that provides them with food, cover for rest and shade, or protection to their nests and scale insects breeding areas (^{Hölldobler and Wilson, 2008}). The foraging area is important to avoid competition for resources among individuals from different colonies. The forage activity of L. apiculatum from March to September presented a peak at 17:00 h, up to 580 m² per colony (^{Ramos et al., 1988}; ^{Mackay and Mackay, 2002}). In this process, the soil vegetation cover -grasses, bushes, agaves, and prickly pear cacti- protects L. apiculatum during high temperature periods (^{Cruz et al., 2014}); but, when vegetation cover is lacking, L. apiculatum has the option to build galleries and tunnels (^{Ramos et al., 1986}).

The pressure on the colonies of ants in most of their distributional areas is related to the demand of this resource. This situation endangers the survival and persistence of this species in Mexico. As a result of this demand and its economic importance during the driest months (February to April) in central-northern Mexico -in addition to the absence of rainfed agriculture-, the extraction of escamoles threatens the survival of ant colonies and their production (^{Ramos et al., 2006}; ^{Esparza et al., 2008}; ^{Tarango, 2012}; ^{Cruz et al., 2014}; ^{Lara et al., 2015}).

Studies on the management and conservation of the distribution areas of the ants are scarce. It is necessary to carry out more research to define the vegetal structure that allows the conservation of the ant, to extend or to improve its habitats and to increase its production. This includes handling domestic livestock, conserving soil and water, and conditioning the extraction of the escamol to a technical standard, that would enable the preservation of the resource for the next generations.

Conclusions

In Villa Gonzalez Ortega, Zacatecas, L. apiculatum mainly nests in A. salmiana and Yucca spp. Each colony frequently carries out its foraging activity with two, three, four, and five trails. The curvilinear and linear foraging distances and the forage substrates vary widely.