Introduction

The vast majority of ant species and genera have unspecialized workers. Monomorphic workers, in contrast to polymorphic ones, are similar in size and shape within the same colony, and as a consequence, their body size distribution is symmetrical and unimodal (Wilson 1953; Holldöbler and Wilson 1990). While polymorphism is associated with division of labor, monomorphism is considered an evolutionary primitive state of colony organization (Oster and Wilson 1978). It is widely assumed that worker polymorphism should enhance colony fitness, but the relationship between colony efficiency and worker diversity is still a matter of debate (Schmid-Hempel 1992; Billick 2002; Fjerdingstad and Crozier 2006).

In some species, the average worker size can differ significantly between colonies despite their being monomorphic (Laskis and Tschinkel 2009; Gouws et al. 2011). For example, the worker head width of Dolichoderus marinae was found to differ between colonies by about 20 % (Laskis and Tschinkel 2009). Such between-colony differences in the worker size can be attributed to particular environmental variations. In the monomorphic Formica aquilonia the head width may differ significantly between the colonies located on clearings and these from forest interiors (Sorvari and Hakkarainen 2009). On the other hand, the head width of queens (gynes) is observed to be relatively insensitive to environmental location or conditions (Haatanen and Sorvari 2013). Furthermore, the values of the within-colony standard deviation of worker size might be comparable among colonies (Laskis and Tschinkel 2009; Sorvari and Hakkarainen 2009).

Theoretically, the average and variance are not the only parameters that can be modified in response to environmental conditions. Other possible changes in the body size may concern the shape of the distribution curve. Two statistics can be used to describe the distribution shape, namely, skewness and kurtosis. Significant skewness indicates an asymmetry of the distribution curve. In turn, kurtosis is the measure of how peaked or flatted the given distribution is compared to a normal one. The value of skewness grows with the increasing frequency of low measurements and shrinks with the increasing frequency of relatively high measurements. Applying both statistics seems desirable when testing for pollution effects on the colony-level traits of ants. Significant skewness indicates a shift between the frequency of big and small individuals. In turn, a significant platykurtic kurtosis suggests a trend towards bimodal distribution (Billick and Carter 2007).

This study concerns the changes in body size of the black garden ant Lasius niger (Formicidae) along a metal-pollution gradient in a post-mining area. The area is contaminated by high concentrations of zinc, cadmium, and lead. L. niger is very common in the investigated area, providing an apt species for testing the effects of chemical pollution on the functioning of ant colonies. We expected that although L. niger is monomorphic, the pollution might significantly affect its morphology. We assumed that the measures describing head size distribution, as well as the measures describing its shape, were prone to modification through the effects of environmental metal pollution.

Materials and methods

Study species and study area

Black garden ants L. niger (L., 1758) are strictly monogynus and monomorphic, building mineral nest mounds containing anything from a hundred to more than ten thousand workers. They inhabit various anthropogenic environments and are widespread in the holoarctic area (Czechowski et al. 2012). Our previous study revealed that L. niger is the most abundant ant, compared to other species inhabiting the investigated area; its relative abundance accounted for more than 70 % of sampled ants (Grześ 2009). The species is both carnivorous and aphidicolous (Czechowski et al. 2012). The transfer of metals to invertebrates is well-documented (Graff et al. 1997; Wilczek et al. 2005; van Straalen et al. 2005; Butovsky 2011; Ardestani and van Gestel 2013; Boshoff et al. 2013; Ding et al. 2013). Similarly, honeydew is considered rich in such metals as Zn, Cd, Pb, and Cu (Stary and Kubiznàkovà 1987). As such, high concentrations of metals in ants, especially of Zn and Cd, were reported by many authors (Migula and Głowacka 1996; Rabitsch 1997; Eeva et al. 2004).

The study area is located in the vicinity of Olkusz in southern Poland. Metal concentrations in the humus layer at the most polluted sites in this region exceed 9,600 mg/kg for zinc, 1,500 mg/kg for lead, and 80 mg/kg for cadmium (Stone et al. 2001). Previous studies performed in the area showed that metal concentrations in the soil were highly correlated with each other (Zygmunt et al. 2006; Stefanowicz et al. 2014). Nineteen sites (S1–S19) were established along the pollution gradient covering abandoned fields and fresh meadows. The transect extended from 0.7 to 35 km from the pollution source. Metal concentrations at the sites decreased as the distance from the smelter increased, approaching background levels at the sites further than 25 km from the smelter. The sites were mostly located on post-agricultural areas (abandoned primarily at the end of the twentieth century), representing formerly cultivated fields and meadows, currently covered by mixed ruderal-meadow-grasslands vegetation. Some plots were also located on the industrial wastelands around the smelter. Well-drained brown loamy or poorly-loamy, neutral to base-rich soils, developed on the shallow sands and the Triassic dolomites, limestone, and the Myślachowice Conglomerate (specific Lower Permian continental calcareous deposits found only on the Kraków-Silesian Upland) dominated the sites (Woch 2011). The study sites (S1–S19) were represented mostly by abandoned fields (14 sites), mown meadows (2 sites), and industrial wastelands (3 sites). In order to describe the study sites, we assessed the habitat type, plant cover, and plant community type on the basis of the botanical data collected during the 2013 vegetation season. For each site, the cover of the vascular plant species was estimated on the five-degree Braun-Blanquet scale (Braun-Blanquet 1964). The Latin plant nomenclature followed Rutkowski (2004). The plant communities of plots were determined according to Chytrý (2007). For each plot, the species richness (total number of plant species), cover, and plant community type were determined.

Pollution and metal analysis

In order to provide a biologically realistic measure of the pollution level of each site, we analyzed Zn and Cd concentrations in a random sample of small invertebrates. The invertebrates were collected at each site by sweep-netting and included mainly Orthopterans, spiders, caterpillars, and beetles. The dry mass of each invertebrate sample accounted for 0.1 g. Total Zn and Cd concentration was analyzed in a preliminary study (Table 1). As both Zn and Cd were found to be highly correlated (P < 0.00005, R 2 = 0.87), we used the total Zn concentration as a single measure of the site pollution level. Additionally, in order to check whether the Zn concentration in the invertebrates reflected the Zn concentration in the soil, three soil samples were collected randomly at each study site, bulked, and analyzed for total Zn and Cd concentration (Table 1). Total concentration of zinc in the invertebrates and soil samples was analyzed using an AAnalyst 800 Spectrometer (PerkinElmer, Boston, MA, USA). Flame atomic absorption spectrometry was applied. Before the analysis, the samples were homogenized and dried at 105 °C. Zinc was extracted with a 4:1 mixture of concentrated Suprapur HNO3 and HClO4 (Baker Instra-Analyzed, JT Baker, Phillipsburg, New Jersey and Ultranal POCH, Avantor Performance Materials Poland S.A., Gliwice, Poland). Two blank samples and two additional reference samples, corresponding to the studied material, were analyzed (sandy loam RTC soil, Sigma-Aldrich, St. Louis, MO and bovine liver CRM 185R, European Commission). The logarithm of Zn concentration in the invertebrates was found to be significantly positively correlated with the logarithm of Zn concentration in the soil (P < 0.005, R 2 = 0.43), meaning that the Zn concentration in the invertebrates reflected the Zn pollution at the study sites.

Table 1 Concentration of total Zn and Cd (milligrams per kilogram) in the upper layer of soil and in random sample of small invertebrates at the study sites (S1–S19) analyzed in the preliminary study
Full size table

Morphological measurements

The species identification followed Czechowski et al. (2012). In each of 19 study sites, 3 mature colonies of L. niger were marked. Each colony was checked for the presence of sexual casts. Six nests were not included in the final analysis due to damage caused by animals or humans or due to a sampling mistake. In total, 51 nests were investigated. A random sample of about 40 workers from each nest was collected and stored at −5 °C. A random subsample of 28 ± 2 workers (apart from one nest where 10 workers were taken) was used for head size measurements. The head size of each ant was measured as the maximum head width above the eyes (“HW” according to Czechowski et al. 2012). Head width is commonly used as convenient index of body size in the number ant species of Formicinae subfamily (Schwander et al. 2005). The Lasius genus head width was previously applied as convenient body size index in Lasius sakagamii males (Yamauchi et al. 2001) as well as in L. niger males and females (Fjerdingstad 2005; Aron et al. 2009). The measurements were performed using a Motic trinocular microscope, model SMZ-168 under a ×30 magnification (Fig. 1). A photograph of each head was taken using a Moticam 1000 digital camera. The head widths were measured digitally to the nearest 0.00001 mm using Motic Imagine Plus 3.0 software (Motic China group, CO., TDL). All measurements were carried out by the same person. The measurements were highly repeatable (r = 0.903, P < 0.0005, based on 30 measurements). As morphological traits in ants may differ between small and big colonies, only mature colonies of big mounds were selected for the purpose of this study, based on their basal area diameter (Holec et al. 2006). In all colonies used in this study, the mound diameter ranged from between 0.4 and 1 m. In L. niger ants, the nest volume grows almost proportionally to the colony size (Rasse and Deneubourg 2001); therefore, the nest basal area diameter was used as an estimator of the colony size.

Fig. 1
figure 1

Measurement of the head width (HW) of Lasius niger using Motic Imagine Plus 3.0 software (Motic China group, CO., TDL)

Full size image

Statistical procedures

To test the relationships between the pollution level and body size of L. niger, the descriptive statistics of head width, including mean, median, coefficient of variation, and range, as well as kurtosis and skewness, were calculated. Each of these statistics was compared separately with the pollution level of the sites (Zn concentration in the invertebrates). The significance of each relationship was tested using a simple regression (Statgraphics Centurion XV) with a 95 % confidence level followed by a check for normality of residuals. The nest size (mound diameter) was found to be significantly positively correlated with the pollution level. Therefore, in order to eliminate the influence of the colony size prior to correlating with the pollution level, each statistic was first correlated with the colony size. The residuals were then regressed against the pollution level. Thus, when testing the relationship between head size and pollution, only the fraction of variance that was not explained by the colony size was used. As three ant nests were collected at each site, the values of residuals for each of the statistics were averaged to avoid a problem of pseudoreplication (Hurlbert 1984).

Results

The study sites, apart from S1, are of xerothermic character, but with diversified plant community types of which the Arrhenatheretum elatioris association was the most frequent. These communities were composed of Arrhenatherum elatius, Festuca rubra, and Leontodon hispidus, with a considerable share of the grassland species, e.g., Galium album, Peucedanum oreoselinum, and Thymus species, as well as some xerophilous ruderal plants (Table 2). In total, 1,482 ants were measured. The head width ranged from between 755 and 1,160 μm. Average head size, median, standard deviation, and range, as well as standardized skewness and kurtosis, were calculated for each nest and correlated with the nest size. Then, the residuals were correlated with the pollution level of the site (Zn concentration in the invertebrates). Average head width differed between the colonies and in some instances by as much as 30 %, and in many cases, the standard deviations did not overlap between the colonies, thereby indicating a strong colony component of variation in the body size. However, no significant relationship was found between the residuals of the average head size and pollution level (P > 0.2). The same was true for the median, coefficient of variation, range, and kurtosis (P > 0.1 in each case). In contrast, the correlation between the standardized skewness and the site pollution was significant (P < 0.005, R 2 = 0.39, Fig. 2). The standardized skewness averaged per site increased with the site pollution from values slightly below zero (−0.5) to 1.5 in the most polluted areas (Fig. 3). This observation indicates that the relationship between big and small ants changed significantly along the pollution gradient.

Table 2 The floristic parameters of the sites
Full size table
Fig. 2
figure 2

Correlation between a standardized skewness of head width in different colonies and the pollution level of the site. The pollution level is expressed as total Zn concentration (milligrams per kilogram) in a random sample of small invertebrates collected at each of 19 study sites. Each point represents averaged colonies within one study site

Full size image
Fig. 3
figure 3

Ideogram illustrating why the size of distribution shape changes along the pollution gradient without significant changes in the mean size. Distributions A and B represent the least and the most polluted site, respectively. If both distributions are skewed in the opposite directions, the averages \( \overline{X_A} \) and \( \overline{X_B} \) are situated close to each other

Full size image

Discussion

The results showed that, although the average and the range of measurements remained unchanged, the fraction of small workers increased along the pollution gradient. In other words, the more polluted the site was, the greater the number of small ants living in the investigated nests.

The explanation why the distribution shape changes along the pollution gradient without significant change in average size is presented in Fig. 4. In the least polluted sites, the distribution of size was found to be slightly biased to the left (distribution A), giving a slightly negative skewness (about −0.5). The average \( \overline{X_A} \), showing a central tendency in the measurements of distribution A, lies below the distribution peak. Distribution B illustrates the distribution of size observed at the most polluted sites. As it tends to be positively skewed, the average \( \overline{X_B} \) lies above the distribution peak. Considering that at the least polluted sites the distribution was only slightly left-skewed, both \( \overline{X_A} \) and \( \overline{X_B} \) are situated close to each other on the x axis. As a consequence, even if the most and the least polluted sites are compared, the difference between averages is never significant and is too low to result in a significant correlation between the pollution level of the site and average values of standardized skewness, obtained for all the investigated sites.

Fig. 4
figure 4

Head size distributions of the ant Lasius niger. A The leased polluted site (S19), B the most polluted site (S14). Each graph represents one nest. For metal concentrations at the sites see Table 1

Full size image

The evidence that the body size in ants has an adaptive meaning is abundant (Davidson 1978; Heinze et al. 2003; Linksvayer and Janssen 2009; Clemencet et al. 2010). Body size and shape can vary in response to environmental constraints such as suboptimal temperature, frequent disturbance, or interspecific competition. For example, in the polymorphic seed-eating ant Veromessor pergandei, the degree of worker polymorphism is associated with co-occurence of ant species with similar forms and diets (Davidson 1978). If the competition pressure was strong, the variability in mandible size of V. pergandei workers decreased due to their specialization in collecting a certain size of seeds (Davidson 1978). Also Powell and Franks (2006) demonstrated a consistent association between a greater variation of prey types and greater morphological diversity in five species of Eciton ants. It is worth emphasizing that under certain conditions, not only the absolute size can be adaptive, but also the range of worker size may enhance colony fitness. This is the case of the weakly polymorphic species Formica neorufibarbis and Formica obscuripes, in which true casts are absent (Billick 2002; Billick and Carter 2007). Similarly, Solenopsis invicta colonies may have great variation in terms of worker length, although discrete worker casts are lacking (Porter and Tschinkel 1985a; Porter and Tschinkel 1985b). The variation of workers size within a colony is important even for the species exhibiting continuous size distribution. However, a coherent theory, explaining in which circumstances social insects should display greater size variation than that expected for the non-social insects, is still lacking (Jandt et al. 2014).

In this study, we demonstrated that the workers of L. niger undergo a subtle change in the head size distribution shape along the metal pollution gradient. As the study was performed within only one polluted area (the vicinity of the Bolesław smelter) and regression analysis was applied, the observed relationship does not evidence of a causal relationship between the observed change and metal pollution level. The effects of other factors correlated with the pollution gradient, but not considered in this study, cannot be ruled out. Such difficulties are inherent in these kinds of studies, when the effects of pollution are investigated on natural populations in the field. However, this study was performed on the meadows representing a range of different plant community types (Table 2), and the relationship between within-colony skewness and pollution was significant even after excluding the effect of the colony size. Thus, the importance of the pollution level in explaining the obtained results should be discussed. Considering the importance of the pollution factor, we propose four hypotheses explaining why the frequency of small ants increases in line with the pollution level.

  1. 1.

    Metal toxicity. Metals in certain concentrations are toxic to living biota, generally due to the inhibiting enzyme activity. A negative effect of Zn and Cd on the growth rate and body size of insects has been shown in a number of studies (e.g., Hare 1992; Fountain and Hopkin 2001; Cervera et al. 2004; Shu et al. 2009). A study performed in the same area as the present study proposed a negative relationship between soil metal contamination and the body mass of adult beetles (Łagisz 2008), although in a similar study, contradictory results were obtained (Zygmunt et al. 2006).

  2. 2.

    Detoxication costs. Ants at the larval stage allocate much energy to body growth. As the ability of the adult ants to filter the contaminated food before feeding the larvae is doubtful (Grześ 2010a), those ants originating from metal-polluted areas are exposed to metal-contaminated food throughout their entire life. It should be noted that the ability to eliminate the excessive amount of metals from the body is quite efficient in some ant species (Grześ 2010b; Gramigni et al. 2013). Nevertheless, maintaining body metal concentration on a certain level should be costly, for example, due to the production of metallothioneins. As a result, the more polluted the site is, the less energy can be invested in in larval growth.

  3. 3.

    Food limitation. Previous works performed in the same area, showed that some taxa of invertebrates decreased in abundance and species diversity along the pollution gradient (e.g., Moroń et al. 2012; Żmudzki and Laskowski 2012). As the investigated species is partly carnivorous, the loss of potential prey diversity and abundance may lead to reduced food sources.

  4. 4.

    Shifts in energy allocation. In ants the adaptation to stress may operate at both individual and colony level (Linksvayer and Janssen 2009). In polluted environments, increased mortality of workers is likely, which could mean the colonies experience a reduction in their foraging force. To maintain an effective number of workers under constrained food supplies, larval development time might become shorter, thereby resulting in a smaller body size of workers. Pollution-induced reduction of life span was documented for another ant species Myrmica rubra (Grześ 2010c). It is also very likely that L. niger colonies lose a considerable number workers at the metal-polluted sites (Grześ et al., unpublished).

    In summary, the stress factors such as metal toxicity and/or limited energy for growth might explain the increasing frequency of small workers in colonies of L. niger in metal-polluted areas. It is worth mentioning that as the within-colony averages remain unrelated to the pollution level, the environmental stressors did not shift the position of the distribution curve on the horizontal axis. As the modification of the distribution shape takes place, the stressors seem to affect the workers within a given colony unevenly. It is likely that depending on a certain genetic background, a specific worker reacts more strongly or weakly to these stressors, and this in turn would lead to the observed bias in the shape of the distribution curve along the investigated gradient.

    This study showed that the distribution shape of the body size of the monomorphic ant L. niger tends to be biased towards a higher frequency of small workers in the more polluted areas. This change may be a result of energy limitation and/or metal toxicity, but also, as suggested by some other studies, it may have an adaptive meaning.