Formicine ants comply with the size‐grain hypothesis

Body size shapes the way organisms perceive the habitat where they live (Levin 1992). For terrestrial walking organisms, as they decrease in size, their environment becomes less planar and more rugose. Smaller organisms perceive the surface of earth as significantly more rugose than large organisms (Morse et al. 1985; Wiens 1989). This was called the size-grain hypothesis by Kaspari & Weiser (1999). As applied to ants, one correlate of that hypothesis is that leg length should increase disproportionately with body mass. Kaspari & Weiser (1999) showed that this is indeed the case with ants, but for one subfamily, Formicinae, in which all linear measures decreased relative to increasing mass. That exception was tentatively explained as an artefact of sampling a reduced range of variation of this subfamily. Here we use a larger database to test whether Formicine ants are truly an exception to the size-grain hypothesis. We have followed exactly the protocol used by Kaspari & Weiser (1999). Six specimens per species were selected to maximize the size range. In dimorphic species only minors were used. Ants were dried for 48 h at 60 °C (Selecta, Martorell, Spain) and weighed en masse on a Sartorius MC1 (Sartorius Corp., Bohemia, NY) microbalance accurate to 0·01 mg. Ants were then glued to points and measured using a Nikon SMZ-U (Nikon Corp., Tokyo, Japan) binocular microscope. Five linear measures were taken for each specimen: maximum head width, not counting eyes (HW), maximum head length (HL), maximum pronotum width, not counting tubercles (PW), hind tibia length (TL), hind femur length (FL). Hind leg length was estimated from (Fl + TL). Measurements were made on 24 species from 16 genera (Table 1), a third of all known genera in this subfamily (Bolton 1995). In four genera (Lasius, Camponotus, Formica, Cataglyphis) different species were used from different subgenera to gauge morphological intrageneric variation. Tables and variables are provided in the same format as those of Kaspari & Weiser (1999). Analyses were made on the mean values for linear measures and mass for the individual species. Allometries were studied using least-squares means regression on log10-transformed values to calculate a and b in the scaling equation y = aMb. To test the null hypothesis of isometry, a t-test on the slope was calculated for H0 = 0·3333 (linear measure vs mass). The same regression analysis was used to find the linear measure that best predicts mass. Mass variation in the 24 Formicine species used varied from 0·035 mg (Brachymyrmex) to 5·40 mg (Camponotus (s. str.)), 2·1 orders of magnitude (Table 1). All linear measures increased with mass (Table 2). Pronotum width (a measure associated with cross-sectional diameter) showed positive allometry with mass, though head width and head length were isometric with mass. Leg length, the variable most pertinent to the size-grain hypothesis, showed that larger Formicine ants are longer-legged than smaller Formicine ants. The null hypothesis of isometry was rejected: the slope of 0·449 was found to be significantly greater (t-test = 2·59; P = 0·008). The size-grain hypothesis also predicts that in small ants leg length is smaller relative to head size. Data on Formicine ants agree with that prediction: log10 hind leg length = 0·388 + 1·256 log10 head length (r2 = 0·86) and the slope is significantly greater than 1 (t-test = 2·4; P = 0·013). The cross-sectional area in larger Formicines is determined by leg size (femur > head length), but in smaller Formicines the reverse applies (head length > femur). Summary statistics of 24 species are given in Table 3. The best predictor of mass is pronotum width (PW) that explains 97% of the variation in worker mass. In the scaling equation y (mass) = aPWb, a = 1·4044 × 10−3 and b = 2·5910. Head length (HL) is the second best predictor for ant mass (r2 = 0·96). For HW r2 = 0·92; for FL r2 = 0·82; for TL r2 = 0·81 and for FL + TL r2 = 0·82. The data set we have studied greatly increases the robustness of the analysis of an important ant subfamily – the Formicinae. All conclusions arrived at by Kaspari & Weiser (1999) are here confirmed for the subfamily Formicinae, reinforcing the general validity of the size-grain hypothesis for the ants, as a group. Specifically (1) leg size increases disproportionately to body mass, with larger ants being relatively long legged, showing a positive allometry for hind leg length (b > 0·33) and (2) leg length is smaller relative to head size in small ants. Pronotum width and head length have also been found to be robust estimators of mass in Formicine ants. Our conclusion, that leg length scales positively to mass, suggests this is the normal condition for all major ant subfamilies. With the data we furnish here, supplementing that provided by Kaspari & Weiser (1999), the size-grain hypothesis is fully applicable over the 3·7 orders of magnitude of dry mass seen in the five subfamilies of ants studied. The size-grain hypothesis deserves further consideration in other taxa of similar size. We thank two referees for careful editing and comments on the manuscript. Numerous colleagues have sent us exotic ants. We must thank specially James C. Trager, Mark B. DuBois and Oleguer Escolà for this. This study as been possible with funds from projects PB91–0482 and PB96–1164 financed by DGES, and with plenty of time from a sabbatical leave (X.E.).