DIVERSITY AND AUTECOLOGY OF WOOD ANTS IN CENTRAL BOSNIA AND HERZEGOVINA (FORMICIDAE: FORMICA S. STR.) Diverzitet i autekologija šumskih mrava u centralnoj Bosni i Hercegovini (Formicidae: Formica s. str.)

Adi Vesnić, Rifat Škrijelj, Sadbera Trožić-Borovac Abstract In this paper new data on diversity of wood ants in Bosnia and Herzegovina are presented. First record of Formica truncorum Fabricius, 1804 from Duboštica is reported for Bosnia and Herzegovina. With this new finding total number of species from subgenus Formica s. str. known in Bosnia and Herzegovina is raised on five. Analysis of wood ants in central Bosnia indicated presence of Formica rufa Linnaeus, 1761 on mountains Trebevic, Ozren, Perun while Formica polyctena Förster, 1850 was collected only on Mt. Čemerska. Morphometrical and ethological intermediary indicate different Formica rufa phenotypes or hybrid Formica rufa x polyctena in investigated area. Polydomy was observed in Formica polyctena and F. rufa. On the mountain Perun 30 colonies of Formica rufa in one kilometer transect were found, which represents the highest known density in Bosnia-Herzegovina.

Wood ants, Formica rufa species group are important ecological component of conifer forest fauna. Colonies of wood ants have up to six million workers and they are important in regulation of plant eating insects (GÖSSWALD, 1981;SEIFERT, 2007). Coniferous woodlands are under strong impact by human activities. Forests of spruce and fir in Bosnia and Herzegovina are primary habitats of Formica rufa and F. polyctena. Survival of wood ants is closely linked to survival of conifer forests. One of one of the major threats to the forest is logging (THOR, 1998). Due continuous coniferous woodland deforestation and ecosystem degradation it is important to collect ecological data, information about distribution of wood ants in Bosnia and 1 Faculty of Natural Sciences and Mathematics Herzegovina. Besides data on the presence of Formica rufa and F. polyctena in Bosnia and Herzegovina no additional information on biology and distribution are known. Complex taxonomy of Formica rufa and F. polyctena is based on number of hairs and cuticle sculpture and does not allow rapid identification (COLLINGWOOD AND AGOSTI, 1986; CZECHOWSKI AND DOUWES, 1996;SOVARI, 2006).
Discrimination of Formica rufa and F. polyctena is difficult due character overlap (SORVARI, 2006). Intraspecific variation character overlap can be result of hybridization between Formica rufa and F. polyctena. In the Europe different phenotypes and even hybrids between Formica rufa and F. polyctena are described (SEIFERT, 1991;SEIFERT, 1992;SEIFERT, 2007). Although the Formica rufa and F. polyctena species were previously reported for Bosnia and Hercegovina there were no data on the possible existence of intermediary phenotypes or hybrid colonies of Formica rufa x polyctena (according to SEIFERT, 2007) or morphs described in earlier studies (SEIFER, 1991). To confirm the existence of different phenotypes in Bosnia and Herzegovina we investigated inter and intraspecific variability in Formica rufa and F. polyctena.
We also evaluated discrimination between Formica rufa and F. polyctena based on subjective and numerical taxonomic characters. Main anthropogenic pressures were evaluated.

MATERIAL AND METHODS -Materijal i metode rada
Investigated area obtained mountains and woodland complexes of spruce and fir in central parts of Bosnia and Herzegovina. Investigations of Duboštica (lat. 44.237 o ; long. 18.313°) were conducted earlier in 2009. Formica trunctorum: 19. 06. 2009; 7 workers. In workers frontal triangle is dull; whool body covered by hairs; second and third antennal scape segments twice as long as broad (Figure 1-2.), identified according to .  During field investigations active methods of collecting were used. In order to analyze structural ecological data the field protocol was applied (AGOSTI ET AL 2000). Data in the field protocol included: locality, sample cod, GPS coordinate, altitude, woodland type (coniferous, deciduous, mixed forest), number of trees around discovered anthill. Number of trees was counted in two perpendicular transects 100 m in length with anthill in center, placement of ant colony in (deep wood, wood edge, meadow, clearings in wood made by construction of wood roads), pressures on habitat (fire, road building infrastructure, wood exploitation, expansion of settlements). Dimensions of anthills for Formica rufa and F. polyctena species were measured: length of longest anthill side (K1), length of shortest anthill side (K2), maximal (R1) and minimal (R2) diameter of colony base. Exposition of an anthill was analyzed.
The protocol was applied on each newly found colony. In laboratory workers were dry mounted on cardboard triangles and reproductive specimens were pinned on entomological needle. Morphometry was performed on Reichert stereo zoom microscope, with micrometer mounted on right ocular. Calibration of ocular was conducted on Reichert. Calibration of microoculare on slide 1 notch = 10 µm. Division of scale in ocular at 100 X magnification was 13,66248 µm and on 50 X magnification 23,529412 µm.
From 01. May till 30. October 2014. we found and sampled 122 nests of Formica rufa and three colonies of Formica polyctena. Workers and reproductive caste were sampled and stored in 96% ethyl alcohol. Five to ten workers were collected form each colony. We analyzed 97 workers of Formica rufa from 32 colonies and 20 workers of Formica polyctena from three colonies.
All findings of colonies identified as hybrid Formica rufa x polyctena or phenotypes of F. rufa. In ecological analysis specimens were treated as Formica rufa .
Statistical analysis included descriptive statistic and inferential ANOVA, post hoc Newman-Keules test. Differences in frequencies were tested by Kruskal Wallis test. Multivariate analysis included discriminant function. Gradient analysis was performed by PCA analysis. Population density was calculated as transect length defined by minimal and maximal altitude of findings divided by number of colonies.  Analysis based on seta number indicated presence of two different phenotypes of Formica rufa in researched area. Typical Formica rufa phenotypes were found in colonies with the head width of worker caste higher than 1.8 mm (Table 7). Typical Formica rufa has the highest average unilateral seta number on occiput, gula and pronotum (Table 7).

RESULTS AND DISCUSSION -Rezultati rada i diskusija
Second Formica rufa phenotype is with significantly (p<5%) narrower head with 1, 7 mm (Tab. 7.). Compared to Formica rufa with typical pilosity second F. rufa phenotype has fewer hairs on pronotum and gula ( Table 7). The fact that ants with smaller head were with less number of hairs is important. There is significant size dependent decline of pilosity in very small workers (SEIFERT, 1991). Juvenile colonies with smaller worker caste exhibit decline of pilosity and were included in analysis. In analysis only five workers were with head width smaller than 1.4 mm. Correlation between head width and number of seta was positive, but linear regression was not statistically significant at 5% level, nPN = -16.60 + 0.01806 * HW, correlation r = 0.50492.   Highest number of seta on pronotum was in workers of Formica rufa. Differences in seta number are statistically significant between Formica rufa, F. rufa phenotype and F. polyctena (Table 5-6).   SEIFERT (2007) did not indicate statistically significant difference in length of seta on gula and pronotum between Formica rufa and assumed hybrid colonies (Graphs 1-2).
Graph 1-2: Distribution of seta number plotted on y axis; gula (left) and pronotum (right) in Formica rufa (dot) and F. rufa phenotype (square) Grafikon 1-2: Distribucija broja seta na y osi; podbradak (lijevo) i pronotum (desno) kod Formica rufa (tačka) and F. rufa fenotip (kvadrat) Two colonies from Perun and one colony from Čavljak show clear characteristics of hybrid Formica rufa x polyctena (SEIFERT, 2007). Determination of hybrid was based on length of seta on gula, pronotum and number of seta on pronotum. Typical characteristics of Formica rufa were found in 42.4% of colonies. Typical Formica polyctena characteristics were found in three colonies.
Number and distribution of seta showed intraspecific and intracolonial variability. Number of worker caste with intermediary seta characteristics between Formica rufa and F. polyctena was 44.5 %. Most significant difference between Formica rufa and F. polyctena was in seta number at gula and pronotum. Average number of seta in colonies of Formica rufa was between 0.0 and maximal 0.7. In colonies of Formica rufa with intermediary phenotypes maximal average number of seta was 0.5 (Table 7).
Microsculpture and glossiness of frons in worker caste proposed by COLLINGWOOD (1979) and AGOSTI AND COLLINGWOOD (1987) was impractical and for Bosnian wood ants characters were poorly defined. Intensity of shines of the frons in samples shows intracolonial variation. Similar data were obtained for seta number and distribution of seta on mesopleuron. Number of seta on mesopleuron also showed intracolonial variability. In same colonies workers without hairs on mesopleuron, workers with hairs on posterior side of mesopleuron and with evenly distributed hairs were found. In worker caste seta on mesopleuron were not detected and it was stabile character in colonies of Formica polyctena.

Statistically significant differences between Formica rufa and hybrids of
Formica rufa x polyctena colonies were detected ( Table 7). Length of scape has statistically significant difference between Formica rufa and F. polyctena (Table 7). Same result was confirmed by scape index (Table 7).  Six variables were included in multivariate discriminant function analysis model. Partial Wilks' Lambda indicates that variable scape length had greatest discriminant power between Formica rufa, F. rufa phenotype and F. polyctena (Table  8). Frontal carina with and scape index did not show statistically significant discriminant power (Table 9).
The first discriminant function is weighted most heavily by the scapus length (SL) and head width (HW). The other two variables also contribute to this function. The second function is marked by head width (HW) and head length (HL) (Table 10; Graph 3-4). The less hairy morph of Formica rufa in our investigations is by hair number very close to F. polyctena. Other authors reported Formica polyctena hairy phenotype from continental Europe suggesting that is hybrid of F. polyctena and F. rufa (SEIFERT, 1991;SEIFERT, 2007;CZECHOWSKI AND DOUWES, 1996). SORVARI (2006) reported Formica polyctena phenotype with more hairs form Finland.   Ecological data between Formica rufa and F. polyctena were not comparable due small number of F. polyctena findings. Statistically significant difference in altitude between Formica polyctena and F.rufa, was detected by multiple comparisons p values Kruskal-Wallis test at 5% significance (Table 11). Differences in anthill dimensions between species were not detected at 5% significance level by Newman-Keuls test. Findings of Formica polyctena were on lower boundary of Formica rufa vertical distribution.
Formica rufa and F. polyctena were most common in coniferous forests. Relative number of colonies of Formica rufa and its phenotype in coniferous Abietopicetum forest habitat was between 88% and 89%. All colonies of Formica polyctena were in Abieto-picetum woodlands. Formica rufa builds colonies more often near wood roads. Total number of Formica rufa colonies near road was 91%, in regard to Formica rufa phenotype 61%. Colonies of Formica polyctena were not found near roads. Data from LUTERŠEK (1960) indicate that Formica polyctena is infrequent, compared to Formica rufa. Distribution data from Igman indicate that large and small wood ant are more common in meadows than in the woods (LUTERŠEK, 1960).
Colonies of Formica polyctena were found in deep woodland habitats 34% and 66% in congregation with more than 30 trees. Colonies of Formica rufa also show strong association with dense forests. Most colonies 86% of Formica rufa were collected in habitats with 20 to 30 trees, similar number of colonies 72% were for Formica rufa phenotype (Graph 5-6). In order to examine association of wood ant to conifer trees we analyzed number of colonies that were built up a tree. In our research we detected that 56% of Formica rufa colonies and 70% of F. rufa phenotype were not leaning up the trees. Colonies of Formica rufa were most commonly built on south, south-east expositions 82%. All colonies of Formica polyctena were built on plane terrain. Multivariate PCA analysis of ecological parameters indicates that first two principal components describe 52% of variability between Formica rufa, F. rufa phenotype and F. polyctena. In construction of first principal component mostly participate variable: habitat type 0.61, altitude -0.51, exposition -0.50. Second principal component is constructed mostly by: habitat density -0.75, construction of colonies up a tree 0.55, position of colonies in regard to road 0.31.
In investigated area wood ants show significant association with dens wood complexes. In closed woodlands number of colonies is higher and polydomy was observed in Formica rufa and F. polyctena. Colonies of Formica rufa on forest edges and fragmented woodland habitats were single. In deep and not fragmented wood complexes colonies were polydome and with large density. In Čavljak area, (Mt. Ozren), with fragmented woodland habitats, number of Formica rufa colonies in transect was 8/1000 m. Highest density of Formica rufa colonies was found on Mt. Perun 30/1000 m. Positive linear correlation between density and number of colonies was detected, but this correlation was not statistically significant at 5% level. Maximal number of Formica rufa colonies decline after its maximum was reached in woodland habitats with 25-30 trees in 100 meter transect. There are two possible explanations for colony number decline in woodlands with higher tree density. In closed woodlands lower temperature is correlated with low insolation level. Second reason was relatively small number of sites with dens woodlands since Formica rufa colonies were dominantly built near wood roads (70%). In our research we found that clearings in forest and wood roads were more attractive for wood ants probably due higher insolation.
In our investigation we detected that woodland roads and clearings in forest have positive effect on distribution and density of Formica rufa colonies. Formica rufa colonies were detected by roads that go through dense forests.

CONCLUSIONS -Zaključci
Central Bosnia is inhabited by five species belonging to subgenus Formica. In narrow sense Formica rufa group in Bosnia and Herzegovina include two species Formica rufa Linnaeus, 1761 and F. polyctena Förster, 1850.
Analysis of seta number in Formica rufa indicates that all samples have low hairiness in compare to central and north European populations. Bosnian wood ants' populations have below average hair number on occiput and pronotum when compared to central and north European populations.
Morphometric analysis of body indicates that Formica rufa less hairy phenotype has much more similarity to Formica rufa than to F. polyctena.