Skip to main content
Dryad logo

Census data of growth, mortality and size stage transitions of nests of leaf-cutting ant species before and after fires


Farji-Brener, Alejandro Gustavo (2021), Census data of growth, mortality and size stage transitions of nests of leaf-cutting ant species before and after fires, Dryad, Dataset,


We hypothesized that fires will affect more the species that build external, easily flammable thatch mounds with superficial fungal gardens (Acromyrmex lobicornis) than colonies that build subterranean nests in the less-flammable bare-ground (Acromyrmex striatus). We use a stochastic matrix demographic model parameterized with 4 years of data in pre and post fire scenarios. This data set contain all the census data which from we build the matrix models.


Field work was performed in a natural reserve of San Luis (“La Florida”), Argentina (33° 07’ S y 66° 03’ W). The area has 340 ha, with an average altitude of 850 amsl (Fig. 1). The average annual temperature in January (summer) is 25 °C and 9 °C in July (winter); the mean annual rainfall is about 600 mm (Del Vitto et al., 1994). The vegetation is represented by species belonging to the Phytogeographic Province of Chaco, Chaqueño Serrano District (Cabrera & Willink, 1980). This nature reserve is occasionally affected by overgrazing, fire, and logging. Due to these disturbances, native plant species typical of Chaco Serrano as well as exotic species are common in the area (Del Vitto et al., 1994).

We worked with Acromyrmex striatus Emery and A. lobicornis Roger, two of the most common leaf-cutting ant species in Argentina in general, and in the study area in particular (Farji-Brener & Ruggiero, 1994). These species are adequate to evaluate the effect of fire regarding nest features. First, these species differ in nest features that may determine the effect of fire on their demography. A. lobicornis constructs nests with an external thatch-mound that may or may not be built at the base of plants (hereafter, “on plants”), and have their fungus garden at the soil surface level inside the mound (Farji-Brener, 2000; Bollazzi et al., 2008). The presence of thatched mounds and their association with plants may facilitate the propagation of fires. Conversely, A. striatus constructs subterranean nests without a mound in areas of bare ground, with multiple chambers excavated up to a depth of several meters where they cultivate the fungal gardens (Boneto, 1959; Diehl-Fleig, 1995; Bollazzi et al., 2008, Fig. 2). Second, in the study area these species live in strong sympatry; nests of A. lobicornis and A. striatus can be found relatively near each other. This fact allows us to easily compare the effect of fire on their survivorship in a similar environment context. Finally, both ant species have a number of traits that facilitate to conduct demographic studies: individual demographic units (i.e., colonies) are easily identified and surveyed at the field, their colonies are founded and maintained by a single queen meaning one can track the demographic history of individual ‘propagules’ from foundation forward, and ecologically relevant proxies for colony size and life-history stage are straightforward to define and measure, which simplifies the construction of demographic models (Fowler et al., 1986; Farji-Brener et al. 2003, Vieira‐Neto et al., 2016).

1. General field methodology

In the spring of 2012, we were randomly walking within the natural reserve looking for nests of both leaf-cutting ant species. We marked 70 nests in total, 35 of Acromyrmex lobicornis and 35 for A. striatus. The sampling includes wide nest sizes to properly perform demographic analyses. Each nest was individually marked and annually censed during the peak of ant activity (spring and summer) during 4 years (2012, 2013, 2014 and 2015). At each visit we recorded the following measures: a) whether the colony was active or inactive (i.e., dead). We considered nests dead if there was an excess of leaf-litter, spider webs, or other debris in the entrances, if no sign of worker activity was detected after disturbing the nest, and if signs of foraging activity were absent (Farji-Brener et al., 2003; Vieira-Neto et al., 2016); b) whether the nest of A. lobicornis was built or not at the base of plants (Fig. 2a and b). As mentioned earlier, the nests of A. striatus is always built on bare ground (Fig. 2c); and c) the size of the nest. Nest size is considered a good estimator of the colony growth in leaf-cutting ants (Fowler, 1977, 1987; Fowler et al., 1986; Vieira-Neto et al., 2016). In the case of the nest-mounds of A. lobicornis, we measured the diameter of the mound because mounds were almost circular in shape (Fig. 2a and b). This nest feature is considered an appropriate estimator of colony growth and had been used in other studies of the same species (Farji-Brener, 2000; Farji-Brener et al. 2003). For A. striatus we counted the number of nest entrances. This parameter is considered a proper estimator for colony growth in leaf-cutting ants nest without mounds (Diehl-Fleig, 1995). After the measurements in 2014, a severe fire occurred in the study area (See Fig. 1). Therefore, to detect the effect of fire on the demography of these leaf-cutting ant species we got measurements from two transition states pre-fire (2012-2013; 2013-2014) and one post-fire (2014-2015).

​​​​​​​2. Population structure and projection matrices

Matrix models are probably the most commonly used in structured population dynamics studies (Caswell, 2001). They are based on two kinds of discretization. On one hand, life cycle of individuals (sensu lato) is subdivided into discrete categories, and on the other hand, its dynamics are described in terms of discrete-time, projecting the population condition from time t to a time t + 1 (for more details of matrix models see appendix 1). In the present work, we evaluate by means of stochastic matrix models the demographic dynamics of A. lobicornis and A. striatus. Our analysis integrates the influence of fire on demographic dynamics according to the species (i.e., nest with/without mounds) and, in the case of A. lobicornis, whether their mounds were built on plants or not. To do so, we calculated the stochastic population growth rate λS (Caswell, 2001) for no fire scenarios and for ones with different fire frequencies.

To build the matrix model and to estimate the demographic parameters, nests were separated in three discrete classes of size: small (1) medium (2) and large (3). In A. lobicornis we followed the criteria used in Farji-Brener et al. (2003). Small nests were those with a mound diameter < 70 cm, medium those between 71-99 cm, and large those > 100 cm of diameter. For A. striatus, we followed the criteria suggested by Diehl-Fleig (1995); small nests were those with 1 or 2 entrances, medium nests those with 3 to 5 entrances, and large nests those with 6 entrances or more. The matrix model is defined by the equation n(t+1) = M(t). n(t), where n(t) is the population vector whose entries give the number (or proportion) of nests in each size class at time t, and M(t) is the population projection matrix (Caswell, 2001). We defined the projection interval (one-time step) as one year. Each matrix entry mij represents the contribution of size class j to size class i for one-time step to the next.  Matrix entries are constituted by two demographic processes: 1) recruitment (i.e. the number of new nests produced by a single queen produced in each nest between one census and the next, i.e., n1, n2, n3); and 2) transition between size classes. The latter includes stasis (the probability of remaining in a class from one census to the next. i.e., a1, a2, a3), growth (reaching another class from one census to the next, i.e., b1à2, b1à3, b2à3) and regression (i.e., moving to a smaller class from one census to the next, i.e., b2à1, b3à1, b3à2). These transitions involve survival and class changing (Fig. 3). 

Recruitment was estimated based on the existing information from leaf-cutting ants in general (Fowler et al. 1986, Farji-Brener et al., 2003; Hölldobler & Wilson, 2011) because there are not data available for the studied species. This kind of estimation has been used in other demographic ant studies (Farji-Brener et al. 2003; Vieira-Neto et al. 2016). Considering that the production of reproductive individuals increases as the colony grows, the higher number of predation in the nuptial flights and the low survival rate of incipient nests (Fowler, 1977, 1987; Vasconcelos & Cherrett, 1995; Farji-Brener et al., 2003; Marti et al., 2015; Vieira-Neto et al., 2016), in both species, we determined a reproductive rate per colony as 100 queens in small nests, 200 in medium nests and 500 queens for large nests with a survival rate of 5%; and a 10% of survival of incipient nests. Therefore, the contributions of new successful nests from one year to the next were defined as 0.5 nests for each small nest, 1 nest for each medium nest, and 2.5 nests for each large nest (Farji-Brener et al., 2003). Despite the subjectivity of these values, their use will not affect the analyses nor the biological interpretations due to the comparative nature of this study. Stasis, growing and regression were calculated from the field data as the proportion of nests that remained in their class, or grew to a higher or decreased to a lower class, respectively.

For all species and nesting types we constructed three projection matrices: two based on field data corresponding to the pre-fire periods (2012-13 and 2013-14), and one after fire (2014-15). After fire, in several cases the observed survival equaled zero. If these matrix entries are given a value of zero, it would effectively imply that mortality is absolute, which is probably not the case. Thus, to solve this and to guarantee projection matrix properties, we followed de Torres Curth et al. (2012) and the observed zero values were replaced by 0.01.

​​​​​​​3. Population projections under environmental variability

After defining the projection matrices, an arbitrary initial population vector was projected and the stochastic population growth rate (lS) was calculated. This stochastic growth rate was obtained from numerical simulations using PopBio R-package (Stubben & Milligan, 2007). The numerical simulation chooses one of the matrices each year and multiplies it by the most recent population vector. This simulation was carried out for four environmental scenarios: no fire, one fire every eight years (last fire record) and for two hypothetical scenarios: one fire every two and four years (fire frequency, FF). To perform the simulations, in non-fire scenario, projection matrices for the periods 2012-13 and 2013-14 get into the simulation process with the same probability (0.5). In the fire scenarios, matrix for the post-fire period (2014-15) enter with probability equal to 1/FF, and those for the pre-fire periods (2012-13 and 2013-14) with equal probability (1-1/FF)/2. The stochastic growth rate (lS) and its confidence interval were obtained for each environmental scenario, species and mound condition (i.e., built or not on plants). Two further analyses were carried out to assess the impact that small changes on demographic processes (i.e., projection matrix entries) have on the stochastic population growth rate: sensitivity and elasticity. Sensitivity analysis measures the impact of small changes in vital rates on population growth rate (i.e., the absolute contribution), while elasticity analysis estimates the effect of a proportional change in the vital rates on population growth rate (i.e., a relative measure of that contribution) (Benton & Grant, 1999). Because elasticities total one, they can be summed in subsets to provide a proportional measure of the importance of each demographic processes for the population growth (Tuljapurkar et al. 2003). Definitions and calculation procedures can be found in Caswell (2001) and Stubben & Milligan (2007).

Usage Notes

This data set is an excell file containing, for two leaf-cutting ant species in pre (years 2012-2014) and pos fire conditions (year 2015) the growth (size stage transitions) and mortality of ant nests during 4 years period.