The system
Pistacia atlantica Desf. (Anacardiaceae) is a dioecious, deciduous tree with a typical Irano-Turanian distribution, from Central Asia through the Middle East to North Africa. Bud burst begins in March and leaf-fall occurs in October. The common and widely distributed aphid, S. betae, induces galls exclusively on leaflets of P. atlantica. The complex lifecycle of the aphid includes sexual and parthenogenetic reproduction, and alternation between a primary host (P. atlantica) and roots of non-specific secondary hosts on which the aphids do not induce galls. Early in the spring, the fundatrix induces small pea-shaped galls (F1 galls), and her offspring crawl out and induce spindle-shaped, “final” galls (F2 galls) on adjacent young leaflets (Fig. 2). Often the density of F2 galls is limited by the availability of young leaflets. The phloem-feeding aphids reproduce in the galls until the fall and then migrate to secondary hosts [37, 45, 53,54,55].
The univoltine (one generation/year) processionary moth, T. solitaria, is the main folivorous insect of Pistacia trees in Israel. Egg hatching is usually synchronous with bud burst in early March. The caterpillars (five instars) feed in dense aggregations on the leaves (Fig. 1) until late May, and then pupate below ground [24, 56]. Occasional population outbreaks of T. solitaria might leave the trees completely defoliated ([24], personal observations). The trees response by rapid compensatory re-growth of young leaves.
Fordini galls are attacked by several natural enemies including unidentified pathogenic fungi, insectivorous birds, a parasitic wasp (M. pistaciaecola), predatory maggots Leucopis sp. (Diptera: Chamamemyiide) and the moth, Alophia combustella. Another moth species (P. guerrini) is a kleptoparasite, destroying the galls by feeding on their inner tissue [26, 53, 57].
Frequencies of tree sharing by the moths and the aphids
Forty and 35 trees were surveyed at Gamla Nature Reserve (32° 54′ N, 35° 44′ E) during May 2006 and 2007, respectively (31 were surveyed in both years). Thirty additional trees were surveyed during May 2007 at Ramot Naftali (33° 6′ N, 35° 33′ E). At this time, the caterpillars had already pupated. On each tree, 20–30 shoots were selected randomly. We recorded earlier feeding by caterpillars (“yes” or “no”) according to the presence of skeletonized leaves (Fig. 1), and counted the number of F2 galls on the trees.
The frequency of host sharing was calculated at both population and tree levels. At the population level: [# shared trees/# surveyed trees] per site. Shared trees had at least one shoot with galls and one (not necessarily the same) shoot infested by moths. At the tree level we calculated: (a) [# shared shoots/# galled shoots]; where shared shoots included all shoots that were infested by both galls (at least one gall) and moths, and (b) [# galls on defoliated shoots/total number of galls].
The differences in host-sharing frequencies between the galls and the moths at Gamla and Ramot Naftali were analyzed by independent sample t-tests. The frequencies of host sharing between years at Gamla (on the same tree) were analyzed by paired t-tests or Wilcoxon signed ranks tests, following Kolmogorov–Smirnov tests for normality. The tests were adjusted with a Bonferroni correction, since data from Gamla were tested twice (between sites and years).
The effects of the moths on the gall-inducing aphids: field experiments
The effects of the moths on the gall density, survival and reproduction of the aphids were examined experimentally on wild populations at Gamla. Galls from defoliated, marked shoots were compared with galls from control (caterpillar-free) shoots that were blocked at their base with Rimifoot, a sticky barrier (RIMI Chemicals Co. Ltd., Israel), that prevents caterpillar access. We examined three gall types: intact (control), trimmed, and those that are induced on the secondary leaf flush (Fig. 2).
The effect of the moths on the distribution and density of F1 and F2
Since galls are induced only on juvenile leaves, their distribution along the shoot provides a reliable time scale of their initiation [44]. We recorded the distribution of the galls (F1 and F2) along the re-growing shoots that were defoliated earlier by T. solitaria. In early June (no more new galls formed), gall distribution was recorded on seven marked trees that were heavily colonized by S. betae. The number of F1 and F2 galls at each leaf position was recorded on 5–7 defoliated, galled shoots, and a similar number of control (blocked) shoots on each tree. The oldest leaf (at the base of the shoot) was marked as leaf #1, and so on. On the defoliated shoots, we distinguished between primary- (often skeletonized) and secondary-flush leaves. The later, younger leaves are softer and reddish in color (Fig. 2).
The effect of the moths on aphid reproduction and attack by natural enemies
In late September, at the peak of aphid reproduction in the galls, the trimmed, intact and secondary leaf-flush F2 galls were collected from seven trees. The galls were opened under a dissecting microscope and the number of aphids within was counted. Only galls that were not attacked by natural enemies were examined. The number of aphids per gall is often highly correlated with gall weight [39]. Therefore, gall weight may provide another assessment of aphid success. The empty trimmed, intact and secondary leaf-flush galls were dried at 70 °C for 48 h and then weighed. In addition, we identified and recorded the rate of parasitism and predation on all gall types (trimmed, intact and secondary leaf-flush).
The leaves of the secondary leaf flush appeared smaller. To test this association, additional secondary leaf-flush and control leaves were collected in October from 5 to 10 defoliated and marked control shoots (respectively) in six of the marked trees. The leaflets from each shoot were counted and then pooled for measurement by a digital leaf-area meter (CI-202, CID Inc., Vancouver, WA, USA).
F1 and F2 gall densities on defoliated vs. control shoots within the same tree were compared separately on primary and secondary leaf flush using paired t-tests. The means of all survival and reproduction parameters between trimmed, intact and secondary leaf-flush galls on the same tree were compared using repeated measures ANOVA (“tree” was the repeated factor) or the Friedman test (depending on the Kolmogorov–Smirnov test for normality). Mean leaflet area per shoot on secondary leaf-flush vs. control leaves on the same tree was compared by a paired t-test. The differences in attack rate by natural enemies between trimmed, intact and secondary leaf-flush galls were tested by a Pearson Chi-Square test for independence.
Moth-feeding experiments
All experiments were conducted with 3rd or 4th instars. Before each feeding experiment, the caterpillars were starved for 24 h.
Feeding on whole galls
The ability of the caterpillars to feed on whole galls, i.e., their ability to deal with the gall’s physical traits (size, thickness, toughness) was examined in non-choice feeding trials. In April, shoots carrying young F2 galls were placed individually in 10 jars with water (average 7 galls/shoot). Five caterpillars were placed on each shoot. Since the caterpillar ate leaves only in this set up, we created artificially defoliated shoots with trimmed galls. To force the caterpillar to eat galls, the leaflets around the galls on these shoots were cut with scissors. The number of damaged or consumed galls was recorded during three consecutive days.
Feeding on ground galls
These experiments aimed at evaluating the chemical component of gall defense against caterpillars. By using ground galls, we eliminated the physical component of gall defense. Young F2 galls and leaves from the same trees were collected during April and ground with a mortar and pestle using liquid nitrogen. The ground galls and leaves were mixed with an artificial caterpillar food (Instant Soybean-Wheat Germ Insect Diet; “Manduca Premix-Heliothis Premix”; Stonefly Industries, Inc.). The diet dough contained leaves or galls in the same relative amount proportion (40% of total food weight). The doughs were placed in Petri dishes with a single caterpillar (10 replicates for each diet). The caterpillars were weighed before and one day after the experiment, and their relative growth rates (RGR) were calculated. In an additional choice experiments, about 0.5 g dough of each diets (galls and leaves) were placed together in each dish with a single caterpillar. The dough was weighed at the beginning of the experiment and for four days (i.e., 24, 48, 72 and 96 h) afterwards. The amount of food consumed of each diet was calculated daily.
In the non-choice experiment, differences in RGR between caterpillars that were grown on a diet with galls compared to a diet with leaves were tested with a Student’s t-test after square-root transformation of the data. In the choice experiment, the proportions of decrease in food weight were compared between gall and leaf diets in the same dishes using a paired t-test after arcsine transformation.
All statistical analyses were conducted with PASW SPSS statistics 17 software. All data were tested for normality prior to statistical analyses using the Kolmogorov–Smirnov test for normality. Only when a normal distribution was found, ANOVA models or t-tests were used. When data did not exhibit a normal distribution, suitable transformations were applied or non-parametric tests were used (as detailed above).