Climate-based model of spatial pattern of the species richness of ants in Georgia
For optimal planning of conservation and monitoring measures, it is important to know the spatial pattern of species richness and especially areas with high species richness. A spatial pattern of the species richness of ants in Georgia (Caucasus) was modeled, areas with the highest number of ant’s species were inferred, and climatic factors that influence the pattern of ant diversity were identified. A database was created by accumulating occurrences for 63 ant species, including 256 localities and 2,018 species/occurrences. Species richness was positively correlated with variables associated with temperature and negatively correlated with variables associated with precipitation. Species richness reaches a maximum at the elevations 800–1,200 m a.s.l. and declines at both lower and higher altitudes. The role of climatic variables and geography of the study area in determining the observed pattern of species richness is discussed.
KeywordsBiodiversity Climatic variables Formicidae Spatial pattern Altitudinal gradient Ground moisture
For optimal planning of conservation and ecological monitoring, it is important to know geographic areas with particularly high species richness (Ceballos and Brown 1995; Garcia 2006; Myers et al. 2000; Newbold et al. 2009). However, development of biodiversity inventories is a very time and effort consuming process, especially for highly diverse species groups such as insects, because it requires intensive sampling and taxonomic expertise (Agosti et al. 2000). An alternative is to develop spatial models based on the limited occurrence data which are already compiled in literature and online databases (Garcia 2006; Newbold et al. 2009).
Understanding spatial pattern of species richness in Georgia is important at the global scale as the whole territory of Georgia is part of the Caucasus biodiversity hotspot, which is one of 25 global biodiversity hotspots. These global biodiversity hotspots comprise only 1.4% of the land surface of the Earth and contain as many as 44% of vascular plant species and more than 30% of vertebrate species (Myers et al. 2000).
Knowing the dependence of species richness on environmental variables is important for understanding the formation of modern species richness distribution and for the prediction of how species richness would be affected by climate change (Kerr 2001; Kienasta et al. 1998; Iverson and Prasad 2001). Dunn et al. (2009) showed that ant species richness is positively correlated with temperature, and negatively correlated with precipitation at a global scale. The use of ants as bioindicators is also growing in Australia (Andersen and Majer 2004), tropical location (e.g. Bestelmeyer and Wiens 2001; Van Hamburg et al. 2004) and temperate areas (Kaspari and Majer 2000; Sauberer et al. 2004).
Taxonomic studies of ants in Georgia started in the late nineteenth century (Gratiashvili and Barjadze 2008). The studies collected much data on the occurrences of ant species (Gratiashvili and Barjadze 2008) but did not identify spatial patterns of the species richness.
Recent algorithms and statistical methods have helped to develop spatial models describing biodiversity including those developed for the prediction of species distributions (Fitzpatrick et al. 2007; Muñoz et al. 2009; Ortega-Huerta and Peterson 2008; Soberon and Peterson 2005; Stockwell 1999; Stockwell and Peters 1999). In order to identify the spatial pattern of species richness, distribution models for single species are developed and then those models are summed (Garcia 2006; Newbold et al. 2009). An alternative method is recording species richness at individual localities and modeling richness patterns directly. Newbold et al. (2009) compared the two approaches while modeling the butterfly and mammal fauna of Egypt. They showed that using the former approach (summing individual models) produces more accurate output. Summing individual models is a good approach only when the available distribution data are sufficient to create individual species distribution models.
In this study, spatial patterns of the species richness of ants in Georgia were modeled and the factors of spatial and altitudinal variation of species richness were inferred.
A database was created by compiling data on the locations of Georgian ant species (Gratiashvili and Barjadze 2008). Geographic coordinates of individual locations were scored from the GeoNames database (http://www.geonames.org/). In total, 2,018 species/locations were analyzed, providing data on 72 ant species and 258 unique localities (Fig. 1).
The modeling of species distribution was performed using openModeller (Muñoz et al. 2009). This software helps to model suitable habitats for individual species and then overlay them in order to estimate summed model of species richness (Muñoz et al. 2009). The GARP algorithm (ecological niche model) was used in order to infer the ant diversity hotspots (Stockwell 1999; Stockwell and Peters 1999).
In total, 19 variables were taken from the WorldClim version 1.4 dataset at a resolution of 30 arcsec (c. 1 km) (Hijmans et al. 2005).These were: (1) annual mean temperature, (2) mean diurnal range, (3) isothermality, (4) temperature seasonality, (5) maximum temperature of warmest month, (6) minimum temperature of coldest month, (7) temperature annual range, (8) mean temperature of wettest quarter, (9) mean temperature of driest quarter, (10) mean temperature of warmest quarter, (11) mean temperature of coldest quarter, (12) annual precipitation, (13) precipitation of wettest month, (14) precipitation of driest month, (15) precipitation seasonality, (16) precipitation of wettest quarter, (17) precipitation of driest quarter, (18) precipitation of warmest quarter, and (19) precipitation of coldest quarter.
Range models were developed for each species with at least five records (Garcia 2006). In total, 72 species met this criteria. Seventy five percent of occurrence locations were used for training the models and 25% was used for validation. Occurrences were divided into test and training points randomly. The accuracy of each model was assessed using the area under the receiver operator (ROC) curve (AUC); the calculations were performed in openModeller with supply of test and training occurrences independently. Following the recommendations made by Swets (1986), nine species with the AUC scores of <0.7 were excluded from the further analysis.
The individual binary models of species ranges predicted by the GARP system were overlaid and summed for producing map of species richness.
Thousand random points were generated using Arcview 3.1, covering whole study area. The following variables were scored for each random point: inferred species richness index, the 19 bioclimatic variables listed above, elevation and Annual Potential Ground Moisture (PGM). Data was extracted from GIS layers using Grid Pig tools extension of ArcView 3.1.
Annual PGM was calculated as: Annual PGM = MP − PET; Where MP = Monthly precipitation, PET = Potential Evapotranspiration; PET = Monthly temperature mean above 0°C × 4.910833333, otherwise = 0 (Thornthwaite 1948; Thornthwaite and Mather 1957).
The regression tree analysis with CHAID (Chi-squared Automatic Interaction Detector, Kass 1980) method was used in order to determine the interaction between species richness and the environmental variables. CHAID analysis is a non-parametric procedure and no assumptions about the data distribution need to be made (Van Diepen and Franses 2006).
SPSS software (SPSS v.16) was used to carry out the analysis. A significance level of 5% was used in the F test, the maximum number of levels was established as three, and the minimum number of cases in a node for being a child node was established at 50. Analysis was performed (1) for each variable separately and (2) including all variables.
Summary table of regression tree analysis using CHAID algorithm for single variables and correlation of species richness with those variables
Regression tree analysis using CHAID algorithm
Mean temperature of driest quarter
Mean temperature of wettest quarter
Temperature annual range
Min temperature of coldest month
Max temperature of warmest month
Mean diurnal range
Precipitation of coldest quarter
Precipitation of warmest quarter
Precipitation of driest quarter
Precipitation of wettest quarter
Precipitation of driest month
Precipitation of wettest month
Mean temperature of coldest quarter
Mean temperature of warmest quarter
Mean annual temperature
Annual potential ground moisture
The regression tree analysis for each variable separately is given in Table 1. Most of the variables were useful for discrimination of species richness except: Precipitation of Coldest Quarter, Precipitation of Driest Quarter and Precipitation of Driest Month. In General Variables associated with precipitation were less useful for discrimination of species richness using single factor. Variables associated with temperature were more predictive for species richness (Table 1).
Variation of species richness with climatic variables
Ant’s species richness, similar to other animal groups, is strongly correlated with climate. Whereas pattern varies by region, the general trend is that richness is positively correlated with temperature and negatively correlated with precipitation and temperature range. This pattern is observed both at a global (Dunn et al. 2007; Dunn et al. 2009) and at a local scale (Wielgoss et al. 2010).
Mean annual precipitation (mm), Annual temperature (°C) and Average species richness
Spatial variation of species richness
Many publications indicate that the number and abundance of species correlates with elevation; Brühl et al. (1999) studied leaf litter ant communities along an elevation gradient on Mount Kinabalu, Malaysia. The number of ant species in the leaf litter decreased exponentially without evidence of a peak in species richness at mid-elevations. Fisher (1999) showed that species richness decreased with elevation linearly in Madagascar ant communities. Samson et al. (1997) surveyed how species richness and abundance in ant communities changes along an elevation gradient in Philippines. Measures of species richness and relative abundances peaked at mid-elevations and declined sharply with increasing elevation. Ants were extremely rare above 1,500 m. Similar patterns were observed in a number of other studies (Olson 1994; Sanders et al. 2003; Sabu et al. 2008). Sanders et al. (2003) surveyed species richness in three canyons in Spring Mountains, Southern Nevada. Ant species richness increased linearly with elevation along two transects and peaked at mid-elevation along a third transect.
Several explanations have been suggested to explain the dependence of species richness on the altitudinal gradient: direct effect of climate (Krebs 2001); indirect effect of climate through net primary productivity—a high net primary production permits consumers to maintain high population densities, thereby reducing the probability of local extinction (Janzen 1973; Siemann 1998; Srivastava and Lawton 1998; Kaspari and Majer 2000); human impact (Sanders et al. 2003).
Annual mean temperature linearly decreases with increase of elevation. Annual precipitation at lower elevations (<700 m) has two basic trends: it either increases or decrease with with a decrease of altitude. This means that low elevation areas in Georgia are either too wet or too dry. After 700 m precipitation varies across mean (900 mm).
The white color in Fig. 8 shows the optimal altitudinal range for ants in space. Even it is not the only factor affecting the spatial pattern of species richness—some basic trends can be observed. The highest species richness (Fig. 8, area 6, marked with quadratic pattern) is concentrated at the biggest and central patch of optimal space But theoptimal spaces which are remote and enclosed by below-optimal space have a lower species richness (Fig. 8, areas 4, 5, 7, 8).
Ant species are very strongly bound within the high elevation mountain ranges of the Great Caucasus from the north and the Lesser Caucasus and arid places in Armenia from the south. Most probably, species turnover within Georgia will be higher than from adjacent territories. Species turnover within Georgia might be limited by the Likhi Range (Fig. 8, area 2) which connects the Great and Lesser Caucasus ranges. This will increase re-colonization times in the western part of optimal space, resulting in lower species richness.
Usage of spatial model of species richness for conservation
Spatial distribution of species richness might be different for other taxa, but many studies have shown that that species numbers of different taxa in the same area correlate with each other (Toranza and Arim 2010; Newbold et al. 2009, Garcia 2006). Therefore, it is logical to expect that species in other groups will show similar spatial distributions to that of ants in Georgia. The area which was shown in this study to have the highest number of ants might be the core area for Georgian biodiversity. Unfortunately, there are no studies on spatial distribution of species richness conducted for other groups in Georgia to test this hypothesis. But, if this turns out to be true, it will be an important outcome for practical conservational. Georgia is still undergoing a process of creation of National Parks. Information about spatial pattern of species richness would be helpful for such planning. Single taxa may not be sufficient for usage as indicators of biodiversity (Oliver et al. 1998) or for detection of overall biodiversity change (Lawton et al. 1998). However, data on species occurrences is accumulated in the literature and in online which can be used for determining important sites for conservation. Distribution data for more than 10 thousand species is available for Georgia (Eliava et al. 2007, Tarkhnishvili et al. 2010), and this study is a first attempt to understand the primary drivers of the spatial pattern of species richness in Georgia. The methods here, can be applied to other taxa and can be used to assist in the decision making processes surrounding where conservation efforts needs to be focused.
I thank Lexo Gavashelishvili and David Tarkhnishvili for providing valuable suggestions during the statistical analysis and for their comments on manuscript. I also express my gratitude to two anonymous reviewers whose comments significantly improved manuscript.
- Agosti D, Majer JD, Alonso LE, Schultz TR (2000) Ants: standard methods for measuring and monitoring biological diversity. Smithsonian Institution Press, Washington, DCGoogle Scholar
- Dunn RR, Sanders NJ, Fitzpatrick MC, Laurent E, Lessard J-P, Agosti D, Andersen AN, Bruhl C, Cerda X, Ellison AM, Fisher BL, Gibb H, Gotelli NJ, Gove A, Guenard B, Janda M, Kaspari M, Longino JT, Majer J, Mcglynn TP, Menke SB, Parr CL, Philpott SM, Pfeiffer M, Retana J, Suarez AV, Vasconcelos HL (2007) Global ant (Hymenoptera: Formicidae) biodiversity and biogeography—a new database and its possibilities. Myrmecological News 10:77–83Google Scholar
- Dunn RR, Sanders NJ, Menke SB, Weiser MD, Fitzpatrick MC, Laurent E, Lessard J-P, Agosti D, Andersen A, Bruhl C, Cerda X, Ellison A, Fisher B, Gibb H, Gotelli H, Gove A, Guénard B, Janda M, Kaspari M, Longino JT, Majer J, McGlynn TP, Menke SB, Parr C, Philpott S, Pfeiffer M, Retana J, Suarez A, Vasconcelos H (2009) Climatic drivers of hemispheric asymmetry in global patterns of ant species richness. Ecol Lett 12:324–333PubMedCrossRefGoogle Scholar
- Eliava I, Cholokava A, Kvavadze E, Bakhtadze G, Bukhnikashvili A (2007) New data on animal biodiversity of Georgia. Bulletin of the Georgian National Academy of Sciences 175(2):115–119Google Scholar
- Fisher BL (1999) Ant diversity patterns along an elevational gradient in the resreve Naturalle Integraled’Andohalela, Madagascar. Fieldiana Zool 94:129–147Google Scholar
- Fitzpatrick MC, Weltzin JF, Sanders NJ, Dunn RR (2007) The biogeography of prediction error: why does the introduced range of the fire ant over-predict its native range? Glob Ecol Biogeogr 16(1):24Google Scholar
- Frouz J, Jilkova V (2008) The effect of ants on soil properties and processes (Hymenoptera: Formicidae). Myrmecol News 11:191–199Google Scholar
- Gratiashvili N, Barjadze Sh (2008) Checklist of the ants (FORMICIDAE LATREILLE, 1809) of Georgia. Proc Instit Zool 23:130–146Google Scholar
- Janzen DH (1973) Sweep samples of tropical foliage insects: effects of seasons, vegetation types, elevation, time of day, and insularity. Ecology 54:687-708Google Scholar
- Kaspari M, Majer JD (2000) Using ants to monitor environmental change. In: Agosti D, Majer JD, Alonso LE, Schultz TR (eds) Ants: standard methods for measuring and monitoring biodiversity. Smithsonian Institution Press, Washington, pp 89–98Google Scholar
- Krebs CJ (2001) Ecology: the experimental analysis of distribution and abundance. Benjamin Cummings, San FranciscoGoogle Scholar
- McCahon TJ, Lockwood JA (1990) Nest architecture and pedoturbation of Formica obscuripes FOREL (Hymenoptera, Formicidae). Pan-Pac Entomol 66:147–156Google Scholar
- Ortega-Huerta MA, Peterson AT (2008) Modeling ecological niches and predicting geographic distributions - a test of six presence-only methods. Rev Mex Biodivers 79:205–216Google Scholar
- Sabu TK, Vineesh PJ, Vinod KV (2008) Diversity of forest litter-inhabiting ants along elevations in the Wayanad region of the Western Ghats. J Insect Sci 8:69. Available online: insectscience.org/8.69Google Scholar
- Siemann E (1998) Experimental tests of effects of plant productivity and diversity on grassland Arthropod diversity. Ecology 79:2057-2070Google Scholar
- Soberon J, Peterson AT (2005) Interpretation of models of fundamental ecological niches and species’ distributional areas. Biodivers Inform 2:1–10Google Scholar
- Srivastava DS, Lawton JH (1998) Why more productive sites have more species: an experimental test of theory using tree-hole communities. Am Nat 152:510–529Google Scholar
- Tarkhnishvili D, Chaladze G, Gavashelishvili L, Javakhishvili Z, Mumladze L (2010) Georgian biodiversity database. Internet: http://www.biodiversity-georgia.net/. Accessed 12 Feb 2011
- Thornthwaite CW, Mather JR (1957) Instructions and tables for computing potential evapotranspiration and the water balance. Publ Climatol 10:311Google Scholar