Revegetation programs are proposed to recover the soil and biodiversity of disturbed sites, this being the case of the Rio Doce basin, Brazil. This region was hugely affected by a mining waste dam disruption, whose leakage on the soil altered its chemical and physical characteristics, and consequently the physiology and performance of plants. The expected alterations of the plants can make them more attractive for leaf-cutting ants, as lower water content induces an increase of non-structural carbohydrates. In this context, we evaluated whether Acromyrmex subterraneus workers differentiate among plants grown on soil with different mining waste concentrations. Leaf discs from plants grown in soil containing 0, 25, 50, 75 and 100% of mining waste were simultaneously offered to ant colonies in a foraging arena. The number of transported discs from each mining waste concentration was recorded until all discs of any concentration had been transported. Leaf selection assays were repeated after 30 days due to the novelty effect phenomenon. Leaf thickness, water, starch and total soluble carbohydrate contents were determined. Leaf discs from plants grown in soil with 100% of mining waste concentration were preferentially selected in both selection assays. Leaf thickness and water content were significantly lower in plants from the aforementioned treatment, while starch and total soluble carbohydrates were higher. Results suggest that seedlings implanted in sites with high mining waste concentration are under high predation risk. Revegetation programs must measure the impact of leaf-cutting ants as both herbivorous and soil ecosystem engineers, for the best management of these insects.

2.1 | Species of study and maintenance of colonies

The experiment was conducted in MirmecoLab at Universidade Federal de Juiz de Fora (UFJF), Brazil, under controlled conditions of temperature and relative humidity (25 ± 1ºC and 75 ± 5% R.H.). We used 10 colonies of A. subterraneus collected in December 2016, and kept in a closed system composed of three compartments interconnected by transparent plastic tubes, corresponding to: foraging arena, fungus chamber and waste chamber. The fungus chamber contained a 1 cm layer of plaster at the bottom to maintain the moisture for the symbiotic fungus. All colonies had approximately 200 cm³ of fungus garden. Acalypha wilkesiana Müll.Arg. (Euphorbiaceae) leaves were provided daily to the colonies. Colonies were leaf-deprived for 24 hours prior to the experimental assays.

2.2 | Plant cultivation and leaf discs preparing

Seeds of Bauhinia rufa (Bong.) Steud. (Fabaceae) from the Plant Physiology Laboratory at UFJF, were germinated in Carolina Soil® substrate, composed by Sphagnum peat moss, expanded vermiculite, dolomitic limestone, agricultural plaster and traces of NPK fertilizer. After germination, 25 seedlings, whose height was 30 cm, were transferred to pots (25 L) with eutrophic substrates at different concentrations. The substrate consisted of a mixture of soil (latosol type, topsoil layer) and sand (1.5:1 by volume). this mixture, it wTo as added a chemical fertilizer, Forth Frutas®, composed by 12% N, 5% P2O5, 15% K2O, 1% Ca, 1% Mg, 5% S, 0.2% Zn, 0.2% Fe, 0.05% Cu, 0.06% B, 0.08% Mn and 0,005% Mo (1 g fertilizer/L substrate) and an acidity corrector, AgroSilício®, composed by 36% CaO and 9% MgO (2 g acidity corrector/L substrate). Mining waste from Fundão dam was then mixed at different concentrations to the eutrophic substrate by volume: 0, 25, 50, 75 and 100%, using a concrete mixer. The mining waste was collected at the Risoleta Neves hydroelectric power plant’s dam, located in the Santa Cruz do Escalvado municipality and stored in a greenhouse at UFJF for two months before the seedlings were planted. About 10.5 million m3 of mining waste from Fundão dam was retained at the reservoir of this hydroelectric (RAMBOLL, 2019). Bauhinia rufa seedlings were kept in the greenhouse since its germination in August 2017 and after 13 months, the 25 individuals were taken for the experimental assays. This species was chosen because it has the characteristics required for revegetation programs (Chaer, Resende, Campello, de Faria, & Boddey, 2011), and it was the one most attacked by leaf-cutting ants in an area where the revegetation program was in progress (P.H.P.Peixoto personal communication, Feb 1st, 2016). 

Leaves from plants grown at different concentrations of mining waste were collected for the experimental assays, and cut in 0.5 cm diameter discs using a metal punch avoiding the leaf’s veins. Each leaf disc was marked with a color corresponding to each concentration of mining waste, using permanent markers (Edding®) (Camargo, Forti, Lopes, Andrade, & Ottati, 2007). Both sides of the discs have been marked to allow their identification when transported.

2.3 | Effect of mining waste concentration on leaf selection

To verify whether the concentration of mining waste in the soil influences the selection of plants, leaf selection assays were performed. Each ant colony was connected by a glass bridge (50 x 3 cm) to a foraging arena containing 10 leaf discs of one plant individual from each concentration of mining waste (total of 50 leaf discs). The discs were randomly arranged in a semicircle form and the assay was started by opening the connection between the colony and the bridge. We recorded how many discs were transported in each of the concentrations of mining waste, one by one, until one of the concentrations had all the discs transported. This procedure was repeated five times in each colony, thus each plant was tested only once by colony, and all plants were tested in all colonies. This constituted the first phase, called "Unknown", of the selection process’ evaluation, which lasted 17 days.

In order to control for novelty effects (Roces & Hölldobler, 1994), we repeated the experiment in a second phase, named "Known", following the same experimental procedures adopted for the Unknown phase. It took place 30 days later and lasted 19 days. The time lapse between the Unknown and Known phases is related to the workers' estimated average lifetime of 90 days (Camargo et al., 2007). During these 30 days, colonies were treated with A. wilkesiana leaves.

2.4 | Leaf analysis

The thickness of the B. rufa leaves was measured from plants kept in the greenhouse with a digital caliper gauge, using 3 fresh leaves from 5 plants of each concentration of mining waste (total of 75 leaves). The thickness of the leaves was determined in points without veins.

The water content of leaves was determined using 150 fresh leaf discs (0.5 cm diameter), being 10 discs from 3 plants of each concentration of mining waste. Leaf discs were immediately weighed, to determine the fresh weight (FW), then oven-dried at 80ºC for 48 hours and weighed again to determine the dry weight (DW). Leaf water content was calculated by taking the difference between the FW and DW, and dividing it by FW (Garnier & Laurent, 1994; Shipley, 1995). 

Total soluble carbohydrates (TSC) and starch content were determined by spectrophotometry, using the method proposed by DuBois, Gilles, Hamilton, Rebers, and Smith (1956) and modified by Shimizu (2009). Thirteen fresh leaves were used, being 2 leaves from 3 plants of each concentration of mining waste. Extractions were performed using samples with 0.05 g FW, homogenized in ethanol 80%. The samples were centrifuged and had their residue separated from the supernatant. From the supernatant, the TSC was measured using the anthrone method (Shimizu, 2009). 

From the TSC extraction residue, the starch content was determined. The TSC extraction residue was re-suspended in perchloric acid (30%) for 30 min, followed by centrifugation. Starch content was also determined by the anthrone method (Shimizu, 2009). In both TSC and starch content analyses, a standard glucose curve was used for calibration and the absorbance was determined at 620 nm.  All extract concentrations were expressed as mg of substance per g of FW.

In the file Leaf_Selection_MiningWasteEffect, column labels refer to: colony: the ID number of each used ant colony; rep: the ID number of each assay; treatment: the mining waste concentration on the soil of the plant from which the leaves were taken; freq: the number of transported leaf discs; moment: the Phases Unknown (A) and Known (B) described in the methods; weight: total number of leaf discs transported in the respective assay.

The file Water_Content_MiningWasteEffect column labels refer to: sample: The ID number of each sample, plant: the ID number of each plant, treatment: the mining waste concentration on the soil of the plant from which the leaves were taken; rep: the ID number of each assay; fresh weight: the leaf disc's weight in g; dry weight: the leaf disc's dry weight in g; water content: fresh weight - dry weight/fresh weight.

The file Leaf_Thickness_MiningWasteEffect column labels refer to: treatment: the mining waste concentration on the soil of the plant from which the leaves were taken; plant: the ID number of each plant; rep: the ID number of each assay; thickness: the leaf disc's thickness in mm.

The file TSC_Starch_MiningWasteEffect column labels refer to: treatment: the mining waste concentration on the soil of the plant from which the leaves were taken; plant: the ID number of each plant; starch: the leaf disc's starch content in mg of starch/g of fresh weight; TSC: the leaf disc's Total Soluble Carbohydrates content in mg of TSC/g of fresh weight.