Introduction
Global climate change, characterized by increased temperatures and temporal variability of precipitation (IPCC, Reference Stocker, Qin, Plattner, Tignor, Allen, Boschung, Nauels, Xia, Bex and Midgley2013), is resulting in a higher frequency and intensity of drought events (Hoerling et al., Reference Hoerling, Eischeid, Perlwitz, Quan, Zhang and Pegion2012). In particular, prolonged summer/autumn drought is predicted to occur more often in Mediterranean regions (IPCC, Reference Solomon, Qin, Manning, Chen, Marquis, Averyt, Tignor and Miller2007). Warming and drought decrease soil water availability and increase atmospheric evaporative demand, leading to growth and physiological changes in water-stressed trees (McDowell et al., Reference McDowell, Pockman, Allen, Breshears, Cobb, Kolb, Plaut, Sperry, West, Williams and Yepez2008). Effectively, in many regions, increased tree mortality is often associated with drought, and sometimes jointly with insect outbreaks (Anderegg et al., Reference Anderegg, Hicke, Fisher, Allen, Aukema, Bentz, Hood, Lichstein, Macalady, McDowell, Pan, Raffa, Sala, Shaw, Stephenson, Tague and Zeppel2015). However, although demographic increase of insect populations can result directly from the positive effects of temperatures on development rates or survival (Ayres & Lombardero, Reference Ayres and Lombardero2000), it is not so straight forward to predict how physiological changes in host trees due to drought may affect insect populations (Jactel et al., Reference Jactel, Petit, Desprez-Loustau, Delzon, Piou, Battisti and Koricheva2012; Anderegg et al., Reference Anderegg, Hicke, Fisher, Allen, Aukema, Bentz, Hood, Lichstein, Macalady, McDowell, Pan, Raffa, Sala, Shaw, Stephenson, Tague and Zeppel2015).
Tree responses to water stress can alter their attractiveness and physical and nutritional value for herbivorous insects (Mattson & Haack, Reference Mattson and Haack1987; Jaleel et al., Reference Jaleel, Manivannan, Wahid, Farooq, Al-Juburi, Somasundaram and Panneerselvam2009). For example, water stress may reduce tree tissue water content and turgor, decrease transpiration and photosynthesis leading to the accumulation of osmolytes such as amino acids and carbohydrates (Rouault et al., Reference Rouault, Candau, Lieutier, Nageleisen, Martin and Warzée2006; Netherer & Schopf, Reference Netherer and Schopf2010) that can increase the performances of herbivorous insects (Mattson & Haack, Reference Mattson and Haack1987; Caldeira et al., Reference Caldeira, Fernandéz, Tomé and Pereira2002). On the other hand, secondary metabolites such as phenols may increase in water-stressed trees, then negatively affecting the performance of herbivorous insects (Koricheva et al., Reference Koricheva, Larsson and Haukioja1998; Huberty & Denno, Reference Huberty and Denno2004; Kolb et al., Reference Kolb, Fettig, Bentz, Stewart, Weed, Hicke, Ayres, Vose, Clark, Luce and Patel-Weynard2016).
Experimental evidence has shown that insect performance may increase (Mattson & Haack, Reference Mattson and Haack1987), decrease (Huberty & Denno, Reference Huberty and Denno2004; Branco et al., Reference Branco, Pereira, Mateus, Tavares and Paiva2010), or remain unaffected (Koricheva et al., Reference Koricheva, Larsson and Haukioja1998; Huberty & Denno, Reference Huberty and Denno2004) in response to water-stressed host trees. Numerous studies have suggested that different insect feeding guilds could show different responses to drought stress, due to different evolutionary and physiological mechanisms (Koricheva et al., Reference Koricheva, Larsson and Haukioja1998; Inbar et al., Reference Inbar, Doostdar and Mayer2001; Huberty & Denno, Reference Huberty and Denno2004; Rouault et al., Reference Rouault, Candau, Lieutier, Nageleisen, Martin and Warzée2006). Generally, bark beetles and wood borers benefit from severe drought (Koricheva et al., Reference Koricheva, Larsson and Haukioja1998; Huberty & Denno, Reference Huberty and Denno2004), sap suckers (e.g., aphids) are impaired (Huberty & Denno, Reference Huberty and Denno2004; Rouault et al., Reference Rouault, Candau, Lieutier, Nageleisen, Martin and Warzée2006), whereas the effect on leaf miners, leaf chewers, and gall makers is more uncertain (Larsson, Reference Larsson1989; Koricheva et al., Reference Koricheva, Larsson and Haukioja1998; Huberty & Denno, Reference Huberty and Denno2004). More intra-specific differentiation and adaptation mechanisms at the population level can modify locally the interactions between herbivorous insects and water-stressed trees. Recent works have highlighted adaptation mechanisms at the population level, due to selective pressures in regard to environmental changes, concerning, for example, the response to temperature (Santos et al., Reference Santos, Paiva, Tavares, Kerdelhué and Branco2011b). Phenotypic traits associated with plant feeding, such as the size of mandibles and their strength (Zovi et al., Reference Zovi, Stastny, Battisti and Larsson2008), or physiological mechanisms allowing insects to cope with plant chemicals (Després et al., Reference Després, David and Gallet2007) may be under strong selective pressure in response to changes in host plants. Within the context of climate change, it is critical to understand if and how insect species are able to cope with drought-stressed hosts (Kolb et al., Reference Kolb, Fettig, Bentz, Stewart, Weed, Hicke, Ayres, Vose, Clark, Luce and Patel-Weynard2016), and if the outcome of such interactions can vary depending on the pest population considered.
Among defoliators, Thaumetopoea pityocampa (Denis & Schiffermuller) (Lepidoptera: Notodontidae), the winter pine processionary moth (PPM) is one of the most damaging defoliators of conifers in Europe, and a threat to human health (Battisti et al., Reference Battisti, Larsson and Roques2017). The species has one generation per year. Over its wide distribution range across the Mediterranean Basin, adult emergence, mating, and egg laying normally occur in the late summer months. Larvae hatch about 1 month later and live gregariously for several months. They feed on pine needles, preferentially the mature ones grown during the previous years (Jacquet et al., Reference Jacquet, Bosc, O'Grady and Jactel2014), throughout the fall and winter. By late winter/early spring, they leave the nest and search for a pupation site in the soil. Adult emergence occurs the following late summer months (Démolin, Reference Démolin1969). This phenology, characterized by the development of larvae in winter, is typical of the PPM throughout its range, and corresponds to the so-called ‘winter populations’ (WP). This typical T. pityocampa phenology is modulated accordingly to climate along its latitudinal and altitudinal range, still keeping summer oviposition. By contrast, a unique population displaying a shifted life cycle, in which adults emerge in the spring and the larvae develop throughout the summer (Santos et al., Reference Santos, Rousselet, Magnoux, Paiva, Branco and Kerdelhué2007) forming the so-called summer population (SP), was discovered in Portugal (in Mata Nacional de Leiria – MNL) ca. 20 years ago. This SP population is genetically differentiated from the local sympatric WP from which it is hypothesized to originate (Santos et al., Reference Santos, Rousselet, Magnoux, Paiva, Branco and Kerdelhué2007, Reference Santos, Burban, Rousselet, Rossi, Branco and Kerdelhué2011a; Burban et al., Reference Burban, Gautier, Leblois, Landes, Santos, Paiva, Branco and Kerdelhué2016). The two sympatric populations are reproductively isolated by time as adults of the two populations have non-overlapping, heritable periods of activity (Branco et al., Reference Branco, Paiva, Santos, Burban and Kerdelhué2017). It represents one of the best documented cases of true allochronic speciation (Taylor & Friesen, Reference Taylor and Friesen2017).
Since its discovery, the SP has been expanding to most parts of the core MNL area (Godefroid et al., Reference Godefroid, Rocha, Santos, Paiva, Burban, Kerdelhué, Branco, Rasplus and Rossi2016). In this region, SP and WP populations co-occur in the same stands, feeding on the same host plants (Pinus pinaster Aiton). In consequence of this shift in the phenology, the different developmental stages of the SP are subjected to different climatic conditions as compared to the WP. In particular, SP larvae develop under high summer temperatures, when trees are generally water stressed, while WP larvae occur in fall and winter, when temperatures are much lower and rainfall is frequent (fig. 1).
Distribution of rainfall and mean monthly temperature in the studied area (meteorological station of Monte Real, Leiria, data for the period 1986–2016). Distribution curves of the young larval stages (L1 to L3) for the two populations, SP and WP, based on the respective males flight period (Santos et al., Reference Santos, Burban, Rousselet, Rossi, Branco and Kerdelhué2011a); the embryonic development (ca. 1 month), and the total period for L1, L2, and L3 development (ca. 2 months).
In this study, we capitalize on the occurrence of the two genetically differentiated populations of PPM, to investigate the potential adaptive response to host mediated-water stress. To achieve this, we first analyzed the effect of drought on needle chemical and physical characteristics, and then compared the performance of the two PPM populations on control and water-stressed host trees during the first larval stages.
We hypothesize that SP larvae can feed more efficiently on tougher needles, and thus perform better than the WP larvae on water-stressed pines. To test our hypothesis we developed an experimental design in order to (i) determine if larval performances were affected by a moderate water stress of the pine host, and (ii) compare the capacity of the two PPM populations to develop on drought-stressed trees, which may underline adaptation mechanisms in the context of climate change scenarios.
Materials and methods
Tree material and treatments
Five-year-old trees of P. pinaster, about 1.8 m high, obtained from a forest nursery (Aliança Florestal, Portugal) were transplanted to pots (volume 25 l, height 35 cm) filled with a substrate of turf, vermiculite 2:1 (v/v), and nutrients (Projar Professional). During 2 months, trees were kept outside, well-watered, and fertilized weekly (NPK 4:3:6 and micronutrients) (Bio-Grow Biobizz) until the experiments started.
Trees were then transferred to a greenhouse and randomly assigned to two different hydric treatments: (i) non-limiting water input (control) and (ii) moderate water stress. Ten replicates per treatment were used. As the larvae of the two T. pityocampa populations develop in different seasons, the feeding experiments were performed accordingly in July-August 2015 for the SP trial and in October–November 2015 for the WP trial. For both trials, the onset of the tree water stress started 3 months before larval hatching (i.e., in April 2015 for the SP trial and July 2015 for the WP trial). Plants were kept in greenhouse conditions: 14/10 h (light/dark), mean air temperature (±SE) 26C ± 2°C, relative humidity 50 ± 2%, during the SP trial; and 12/12 h (light/dark), 23 ± 2°C and 60 ± 2% during WP trial.
Water stress was imposed through the incomplete replacement of the water transpired from the plant, which was estimated by weighting the tree pots weekly. Replacement of the water lost was achieved by successively reducing this amount by 10% per week, until moderate water stress was attained as defined by the leaf water potential, and then maintained at that level. Water from the pots was only lost through tree transpiration as the top of the pots was closed with a white plastic sheet to avoid direct evaporation from the soil. Well-watered trees were watered to soil field capacity once or twice per week, depending on air temperature. Air temperature and relative humidity (reference sensor) inside the greenhouse were continuously monitored with an Em50 datalogger (Decagon Devices Inc, Washington, USA).
Plant water status
In both SP and WP trials, tree water condition was monitored by measuring the predawn needle water potential (ψpd) in a pair of 1-year-old needles (i.e., grown the previous year) for all trees, using a Scholander pressure chamber (PMS-1000, PMS Instruments, Corvalis, Oregon, USA). The water potential was evaluated twice for each trial, 1 month previously to the beginning of the test and at the start of the feeding experiments. Tree water status was also determined by measuring the needles relative water content (RWC), a parameter that reflects the metabolic activity in tissues and is used as an index for dehydration tolerance (Anjum et al., Reference Anjum, Xie, Wang, Saleem, Man and Lei2011). Ten needles per tree were collected at dawn and immediately weighted to obtain their fresh weight (FW); next they were placed in distilled water for 24 h in the dark, to achieve turgid weight (TW) and finally samples were dried at 80°C for 48 h to obtain their dry weight (DW). RWC values were calculated as: RWC = (FW − DW)/(TW − DW) × 100.
Foliar analyses
We assessed relevant needle biochemical (chlorophyll, soluble sugars, phenols, carbon, and nitrogen), and physical (hardness) characteristics using 1-year-old fresh needles from controlled and stressed trees from SP and WP trials. The needles for chemical analysis were collected when feeding trials started, and were stored at −80°C. For chlorophyll and soluble sugars, three samples from each tree were analyzed from different tree canopy expositions, while for phenols, carbon, and nitrogen, one composite sample per tree, comprising needles collected from different tree orientations was used. Still, in both cases, representative samples were obtained for each tree.
Chlorophyll content (a, b, and total) was determined using a modified version of Arnon (Reference Arnon1949) method. Extractions were made from 0.1 g of fresh material with ethanol 80% (v/v). After centrifugation at 13,000 rpm for 5 min, absorbance was read at 663 and 645 nm for chlorophyll a and b, respectively. Soluble sugars (glucose, fructose, and sucrose) were measured as described by Stitt et al. (Reference Stitt, Lilley, Gerhardt and Heldt1989) based on the enzymatic method of Jones et al. (Reference Jones, Outlaw and Lowry1977) using 0.1 g of fresh needles. Total phenols content were determined based on the Folin–Ciocalteu method (Folin & Ciocalteu, Reference Folin and Ciocalteu1927) using 0.5 g of dried (48 h at 60°C) fine powered needle. Calibration curves were created using standard gallic acid solutions and the amount of total phenols was expressed in mg of gallic acid equivalents per g of dry weight.
Total needle nitrogen concentration was analyzed by a modified Kjeldahl analysis (Berthelot reaction, modified) (Searle, Reference Searle1984) using 0.5 g of fine-milled needles. Samples were digested with 5 ml sulfuric acid (98%) and selenium as a catalyst for 3 h at about 350°C. On cooling, it was diluted to 50 ml with distilled water and shaken thoroughly. Quantifications were made by molecular absorption spectrophotometry in the visible region (660 nm). The N-concentration of the digests was determined on a continuous flow analyzer (Skalar SanPlus System, Breda, The Netherlands), while total needle carbon (C) was determined by quantification of CO2 by infrared after combustion at 1200°C (Analytik Jena EA 4000, Jena, Germany). Results were expressed as a percentage of dry matter in both cases, and the C/N ratio was determined.
Needle toughness was measured with a penetrometer that determined the strength necessary to rip the needle epidermis, as an indicator of the strength that larvae would need when feeding on the needles. Ten pairs (1-year-old) needles per tree for each plant treatment and trial (SP and WP), were measured for toughness in the middle of the needle top surface. Values were expressed in units of force (Newton).
PPM population feeding trials
Larvae of the two PPM populations were obtained from egg masses collected in the field, from P. pinaster trees. For SP populations, egg masses were collected at the end of June, in Nazaré (39°36′50.70″N; 9°04′25.80″W), and for WP, collections were made at the end of September, in Setúbal Península (38°34′42″N; 9°07′35″W). Egg masses were individually placed in glass test tubes and kept at room temperature (22–28°C) until larval hatching. For both populations, 30 newly hatched larvae were used in the trials that started at the beginning of July for SP and at the end of September for WP. The feeding experiments were carried out in glass Petri dishes (140 mm in diameter) where larvae were fed, following the feeding protocol described in Hódar et al. (Reference Hódar, Zamora and Castro2002). Each Petri dish corresponded to one tree of each treatment. In total, each PPM population feeding trial consisted of ten replicates per treatment. At each time, four pairs of needles (1-year-old), inserted into the wet floral foam to avoid excessive loss of water, were offered to the larvae. The Petri dishes were placed in rearing chambers with controlled temperature (25 ± 2°C) and relative humidity (60 ± 2%) within the range of optimal conditions for larval development (Démolin, Reference Démolin1969), at the entomological laboratory of Forest Research Centre, Lisbon University. Larvae were inspected every 3–4 days for food replacement, cleaning, feces removal and counting of dead larvae. Larval performance was estimated by larval survival, consumption rate, feces production and development time (number of days to reach the 2nd instar) for each treatment. The development time until 3rd instar was not estimated due to the low number of individuals reaching this stage, especially on the WP trial. The consumption rate was determined by measuring the total length (mm) of needles consumed. Feces production was calculated based on the feces dry weight. Both consumption and feces produced were then divided by the number of individuals alive at each sampling date, to obtain average values per individual. The feeding trials were conducted until larvae reached the 3rd instar, after which they become urticating and thus difficult to handle.
Statistical analyses
Independent sample t-tests were performed to compare needle water potential between treatments (stressed and control), for each date of measurement. Levene's test was used to test for equal variances between groups. Whenever equal variances were not assumed, degrees of freedom were adjusted. RWC, chlorophylls, soluble sugars, and toughness were analyzed through Generalized Estimating Equations (GEE), with normal distribution and identity link function. Each tree was a replicate and needles from the same tree were repeated measurements accounting for within-subject effect. Consumption and feces production by the larvae were also analyzed through GEE, with normal distribution and log link function. Each rearing Petri dish was a replicate and observations on different dates were considered repeated measurements. Generalized linear models (GLM) with log link function were used to test for differences between treatments of C/N values and larvae development time. A two-way ANOVA was used for analyzing the effects on total phenols. For all models, the factors tree treatment and trial were considered as categorical predictable variables. Linearly independent pairwise comparisons among the estimated marginal means were conducted using Least Significant Difference test. For GLM and GEE models, statistics are presented in the form of Wald χ2. Degrees of freedom were 1 in all cases. The mean number of days of larval survival for each trial was estimated by the Kaplan–Maier method. Individuals alive at the end of the period were used as censored ones. The Log Rank (Mantel–Cox) test was used to compare mean survival between groups of larvae from different treatments. Statistical analyses were conducted with IBM SPSS statistics 23 software.
Results
Tree parameters
Two months after the stress treatment started, that is 1 month before the beginning of the feeding trials, the needle predawn water potential (Ψpd) was already significantly lower in stressed than in control trees, for both SP and WP trials (fig. 2) (Trial SP: T-test = −7.344, df = 10.52, P < 0.001; Trial WP: T-test = −5.120, df = 9.98, P < 0.001). Differences increased until the onset of the feeding trial (0 month) (fig. 2) (Trial SP: T-test = −12.489, df = 18, P < 0.001; Trial WP: T-test = −6.873 df = 9.79, P < 0.001). Levene's test further evidenced unequal variance with higher variance on the stressed trees in comparison with control ones for most cases.
Predawn needle water potential (Ψpd, MPa), (mean ± SE), on drought-stressed and control trees, of trials SP and WP, measured 1 month and just (0 month) before the larvae feeding experiment started.
Accordingly, needles RWC was higher for control (Trial SP = 94.4 ± 0.25%; Trial WP = 92.7 ± 0.17%) than for stressed trees (Trial SP = 80.9 ± 0.16%; Trial WP = 77.6 ± 0.52%). Differences between tree treatments, considering both trials, were statistically significant (Wald χ2 = 2107.193, df = 1, P < 0.001).
Tree treatment produced significant effects on all biochemical compounds analyzed, as well as on needles toughness (tables 1 and 2). Consistently, total soluble sugars, phenols, and toughness were higher in water-stressed trees, in both trials. An opposite trend was found for chlorophylls, which were higher on control trees than on stressed ones (table 2). The effect of the trial (SP or WP) was only significant for the carbon/nitrogen ratio and not for the remaining parameters (table 1). C/N was higher on stressed trees but differences were only significant for the SP trial. The interaction trial × treatment was never significant (table 1). Total soluble sugars were represented mainly by sucrose accounting for 67 and 74% in control and stressed trees, respectively. Overall, glucose accounted for only 6% and fructose to 23.5% of total soluble sugars content.
Results of statistical analysis on needles' traits from two treatments (stressed and control trees) and two trials (SP and WP).
For C/N, sugars, Chlorophylls and Toughness, statistical test corresponds to Wald χ2 (df = 1) obtained with Generalized linear models; for phenols, a two-way ANOVA was used (F 1,36). For all models, the factors treatment (Treat.), trial, and its interaction term were considered. In italic P-values <0.05.
Physical and chemical parameters analyzed on stressed and control trees and on SP and WP trials (mean ± standard error).
Different letters mean statistically significant differences between treatments determined by multiple pairwise comparisons through Least Significant Difference (α = 0.05).
Larval parameters
Log Rank (Mantel–Cox) test showed an overall significantly lower survival of larvae fed with stressed trees in comparison with control ones (χ2 = 71.837, df = 1, P < 0.001). Still, pairwise comparisons (i.e., stress vs. control groups) for each PPM population evidenced significant differences only for WP, with the survival of the larvae fed on stressed trees decreasing sharply as from the first days of feeding (fig. 3). By the end of the trial, mortality reached almost 100% for larvae fed on stressed trees, but was also very high for the control ones. Kaplan–Meyer estimates for the average number of days of survival for SP larvae were 21.1 ± 0.7 and 19.6 ± 0.6 days, respectively, for larvae fed on control and stressed trees. However, estimated survival (expressed as the number of days) was lower for the WP larvae than for the SP, especially for those fed on needles from stressed host trees (18.8 ± 0.7 and 11.5 ± 0.5 days, on control and stressed hosts, respectively).
Larvae survival estimates (±SE) of the two phenotypic divergent populations of T. pityocampa (SP and WP) reared with pine needles of drought stress and control trees.
The mean development time needed to reach the 2nd instar was shorter for larvae fed on control trees (12.3 days ± 0.5) than for larvae fed on stressed trees (13.8 days ± 0.6). However, differences were not significant (Wald χ2 = 3.685, P = 0.055). Similarly, differences between PPM population feeding trials were borderline (Wald χ2 = 3.630, P = 0.057), with larval development being on average faster in SP than WP. The interaction term was not significant (Wald χ2 = 0.193, P = 0.660).
Tree stress had an overall significant negative effect on the needle consumption (Wald χ2 = 4.439, P = 0.035). Overall, average consumption was 5.48 ± 0.54 mm per larvae.day−1 in the control group and 4.00 ± 0.45 mm on the stressed plants (fig. 4). On average, there were no differences between the SP and WP trials (Wald χ2 = 0.121, P = 0.728) and the interaction term plant treatment x trial was not significant (Wald χ2 = 1.328, P = 0.249). Yet, pairwise comparisons between the control group and stressed plants showed that differences were higher for WP (P = 0.072) than for SP (P = 0.256). Concerning the production of feces, no significant effect was found for treatment (Wald χ2 = 1.731, P = 0.188), population (Wald χ2 = 0.003, P = 0.957), or the interaction term (Wald χ2 = 2.326, P = 0.127).
Consumption (in millimeters) and production of feces (in milligrams) by the larvae of both trials (SP and WP) on drought-stressed and control trees (mean ± SE). Different letters mean statistically significant differences between treatments determined by multiple pairwise comparisons through Least Significant Difference (α = 0.1 and α = 0.05 – indicated with *).
Discussion
Overall, we observed a tendency for a negative effect of plant water stress on larval performance, reflected on a lower larval survival, slower development, and lower consumption rate. Yet, a negative effect was significant only for the WP, mainly resulting in a lower larval survival. SP larvae survived, on average, 2 and 8 more days than WP larvae when fed on control and stressed plants, respectively. Taking into account the shifted phenology and the recent and local origin of the SP (Leblois et al., Reference Leblois, Gautier, Rohfritsch, Foucaud, Burban, Galan, Loiseau, Saune, Branco, Gharbi, Vitalis and Kerdelhue2017), it can be considered that SP larvae have been adapting to summer dryer conditions.
The negative effect on the performance of the larvae may be explained by changes in plant chemical and physical traits. Water-stressed plants registered higher amounts of soluble sugars, higher C/N ratio, and phenols; needles were tougher, whereas water content and chlorophylls concentrations were lower. Water-stressed trees reached similar stress values and chemical traits for SP and WP trials, which suggest that the differences found in insect performance are not due to a difference in plant treatment. The leaf water potential of the stressed trees in both trials was in the range of values measured in P. pinaster trees under field conditions during the summer (e.g., Loustau et al., Reference Loustau, Berbigier, Roumagnac, Arruda-Pacheco, David, Ferreira, Pereira and Tavares1996).
Most of the results obtained in this work are supported by literature references. Several studies document effects of drought stress in plants reflected in leaf water potential, RWC, turgor, transpiration rate, and photosynthesis (Huberty & Denno, Reference Huberty and Denno2004; Jaleel et al., Reference Jaleel, Manivannan, Wahid, Farooq, Al-Juburi, Somasundaram and Panneerselvam2009). Consequently, many components of plant nutritional quality are modified with possible effects, either positive or negative, on the performance of associated insects. The outcome may also depend on how insects feed on plant tissues (Huberty & Denno, Reference Huberty and Denno2004). Several studies concluded that drought often increases the concentration of sugars and inorganic ions, as well as nitrogen compounds, such as amino acids, in pant tissues thus enhancing foliage nutritional quality for some herbivore insects (Huberty & Denno, Reference Huberty and Denno2004; Rouault et al., Reference Rouault, Candau, Lieutier, Nageleisen, Martin and Warzée2006; Netherer & Schopf, Reference Netherer and Schopf2010; Jactel et al., Reference Jactel, Petit, Desprez-Loustau, Delzon, Piou, Battisti and Koricheva2012). Yet, in the system here studied, the possible positive effect of increased sugar concentration was insufficient to counteract the negative effects of plant drought stress on larvae. At the same time, our results confirmed that a moderate water stress applied to P. pinaster trees led to a significant decrease in chlorophyll concentration and higher C/N ratio, similarly observed by Chen et al. (Reference Chen, Wang, Xiong, Cao and Deng2015) for sorghum seedlings.
On the other hand, a negative effect on insect performance can occur due to a decrease in leaf water content, increased leaf toughness or to an increase of secondary metabolites concentration. According to Henriksson et al. (Reference Henriksson, Haukioja, Ossipov, Ossipova, Sillanpää, Kapari and Pihlaja2003), leaf water content is one of the most important factors influencing larval growth of Epirrita autumnata (Lepidoptera). In our study, stressed plants revealed lower values of leaf water potential and RWC than control ones which could explain in part, the negative effects observed on larval performance. A compensation mechanism was not observed for PPM, as consumption was lower when feeding upon needles of stressed plants than from non-stressed ones. We also found that stressed needles were tougher than control ones, which could explain the lower larval consumption observed in water-stressed plants. Interestingly, a previous study (Santos et al., Reference Santos, Paiva, Rocha, Kerdelhué and Branco2013) showed that SP eggs are generally bigger that WP eggs, suggesting that SP neonate larvae are also bigger, with stronger mandibles. This trait could allow SP young larvae to feed more efficiently on tougher needles, and thus perform better than WP larvae on water-stressed pines. This could explain the significant difference in survival observed, since the first days when neonate larvae were present (fig. 3). A similar hypothesis was suggested by Zovi et al. (Reference Zovi, Stastny, Battisti and Larsson2008) who documented a case of local adaptation of T. pityocampa larvae to soft vs. tough needles. It is worth noting that overall mortality was very high by the end of the feeding trial, even in the control groups, which is explained by the high sensitivity of this insect species to laboratory rearing conditions during prolonged periods (Berardi et al., Reference Berardi, Branco, Paiva, Santos and Battisti2015; Branco et al., Reference Branco, Paiva, Santos, Burban and Kerdelhué2017). The production of feces was not affected by plant treatments. This particular outcome, lower consumption but unchanged feces production, suggests a reduction in digestibility of the pine needles following water stress. We could not calculate digestibility indices to confirm this hypothesis due to the gregarious behavior of this insect.
Concentrations of several classes of defensive secondary compounds such as cyanogenic glycosides, phenols, terpenoids, and alkaloids may also increase in plant tissues in response to moderate drought stress (Mattson & Haack, Reference Mattson and Haack1987; Kolb et al., Reference Kolb, Fettig, Bentz, Stewart, Weed, Hicke, Ayres, Vose, Clark, Luce and Patel-Weynard2016). In both trials, total phenols were significantly higher in water-stressed plants compared to control ones which might have further accounted for the negative effect on larvae, especially on their survival. A higher detoxifying capacity of SP vs. WP larvae could have occurred thus explaining that the negative effect of stress was only significant for the WP. This hypothesis could be explored via dedicated experiments and analyses of differential expression of detoxification genes in both populations in stress vs. control groups.
The overall better performance of SP larvae on stressed plants by comparison to WP may indicate an evolutionary adaptation to cope with the changes resulting from host plant drought stress. In fact SP larvae feed during summer months, facing environmental conditions such as high temperatures, low air humidity, and low rainfall (fig. 1), all contributing to plant drought stress, whereas WP larvae feed during fall and winter, when the tree's physiological parameters are supposed to be less stressed. Previous studies demonstrated that the SP has already undergone some ecological divergences in relation to the WP, namely regarding the upper thermal tolerance limits of the larvae (Santos et al., Reference Santos, Paiva, Tavares, Kerdelhué and Branco2011b) and of the eggs (Rocha et al., Reference Rocha, Kerdelhue, Ben Jamaa, Dhahri, Burban and Branco2017), as well as some reproductive traits, such as fecundity and egg size (Santos et al., Reference Santos, Paiva, Rocha, Kerdelhué and Branco2013). Also, SP larvae had a faster development than WP, with more larvae reaching the 3rd instar in both treatments. Overall, results suggest that the SP is better adapted to develop on stressed hosts and might have a competitive advantage over the WP in some specific climatic conditions.
Conclusions
Under present climate change scenarios, characterized by rising temperatures and moderate to severe drought stress episodes, plant drought stress will accordingly increase and modulate plant–insect interactions. This work proved that water stress affected negatively the performance of young larvae of the pine defoliator species studied. A major finding revealed that the SP population, with summer larval feeding, was less affected by stressed trees than the WP population, indicating that some adaptation to the physiological status of the host plant has occurred following the phenological shift. This could be due to stronger mandibles, ability to detoxify secondary compounds, and/or capacity to withstand feeding upon low water contents foliage. The present study also highlights the relevance to understand evolutionary and adaptation mechanisms that might occur in the context of global climate change and how this will affect species predictions under different scenarios.
Acknowledgements
We thank Dr Maria Rosa Paiva, FCT-UNL for a critical reading of the text. We also thank Catarina Tavares and Raquel Portanova for help with water potential measurements and Ana Rodrigues, Miguel Martins and Joaquina Silva for technical laboratorial support. Special thanks to Joaquim Mendes for his always prompt availability to help with whatever needed. This study was supported by Susana Rocha PhD scholarship with the reference SFRH/BD/90188/2012(FCT-MCES) and CEF (Centro de Estudos Florestais). CEF is a research unit funded by Fundação para a Ciência e a Tecnologia I.P. (FCT), Portugal (UID/AGR/00239/2013).
