Trophic impacts of a rolled‐rye cover crop on the lettuce aphid (Nasonovia ribisnigri) and associated insect fauna in cultivated Histosols

Cultivated peatland (Histosol) in Southern Québec (Canada) is a rapidly declining non‐renewable resource used to grow most Canadian lettuce (Lactuca sativa L., Asteraceae). Rolled‐rye (Secale cereale L., Poaceae) cover crop is one of the conservation practices proposed to reach a more sustainable lettuce production, but the overall impact on the agroecosystem remains poorly studied in Histosols. We assessed multiple effects of rolled‐rye cover crop on the trophic chain associated with the lettuce aphid, Nasonovia ribisnigri (Mosley) (Hemiptera: Aphididae), a major pest of lettuce. During one growing season and through two consecutive lettuce crops, we monitored in situ the impacts of rolled‐rye cover crop on insect fauna and lettuce quality. We used visual scouting and yellow pan traps to assess plant colonization by N. ribisnigri, its natural enemies, and alternative prey. Exclusion cage experiments were also conducted to measure aphid fitness and population growth. Under greenhouse conditions, following cover crop removal, we examined potential lingering effects of rye within the soil on lettuce plants and N. ribisnigri. In situ experiments showed that rolled‐rye cover crop has the potential to inhibit N. ribisnigri field colonization and recruit natural enemies and alternative prey in the first lettuce crop. Rye also reduced aphid fitness as well as lettuce foliar amino acid concentration and weight. For the second lettuce crop, an almost 50% reduction in N. ribisnigri abundance was observed with the use of rolled‐rye cover crop. In greenhouse experiments, no persistent effect of rye was observed on the quality of lettuce grown with soil collected under a cover crop, nor on the fitness of aphids inoculated on these lettuces. This study highlights the diversity of trophic effects rolled‐rye cover crop may have on lettuce production in Histosols and the potential of rolled‐rye cover crop as a cultural practice to reduce aphid populations. Mechanisms at play while underlining agronomic challenges regarding proper rye termination must be further explored to maintain high‐performing lettuce yields.

Beside rolled-rye cover crop's potential as a soil protection measure, its impacts could cascade through the trophic chain from soil microbiota to lettuce and arthropod fauna, including herbivorous pests and their natural enemies (Kabouw et al., 2011;Kim et al., 2020).Aphids, through high parthenogenetic reproductive rate and adaptive polymorphism (dispersing winged and reproductive apterous individuals) are common pests to many crops (van Emden & Harrington, 2017).Amongst the 21 aphid species known to feed on lettuce, the lettuce aphid, Nasonovia ribisnigri (Mosley) (Hemiptera: Aphididae), is the most economically important (Forbes & MacKenzie, 1982;McCreight, 2008).Infesting young leaves and heads of lettuce plants, it can cause severe cosmetic damage (Morales et al., 2013).Foliar insecticides often remain ineffective because of the low penetration of the chemicals into the heart of the lettuce (ten Broeke et al., 2013).Biological control practices such as flower strips for natural enemy recruitment show promise against N. ribisnigri but are rarely employed outside of organic productions (Brennan, 2013).
As cereal cover crops have been shown to increase the control of Aphis glycines Matsumura on soybean (Koch et al., 2012(Koch et al., , 2015) ) and Myzus persicae (Sulzer) on cabbage (Bottenberg et al., 1997), rolled-rye cover crop could represent an interesting avenue for controlling N. ribisnigri in lettuce.Rolled-rye cover crop may contribute to recruiting aphid natural enemies (Schmidt et al., 2004) by creating new refuges characterized by increased structural complexity and cooler microhabitats that attract diversified arthropod communities (Landis et al., 2000).The disruption of a well-defined visual crop-soil contrast created by the cover crop could also impede aphid orientation and reduce field colonization by alates (Costello, 1995;Saucke & Döring, 2004).Rolled-rye cover crops are also known to reduce weed emergence through allelopathic interactions, competition for water and nutrients, and physical obstruction (Teasdale et al., 2012).Plant-soil interactions are also affected because, alongside temperature and humidity, the high C:N ratio of rye litter leads to nitrogen immobilization and modified soil microbiota (Kos et al., 2015).Modified soil nitrogen availability and plant metabolism can in turn affect plant-aphid interactions as minute dietary changes can affect aphid fitness (Douglas, 1993).Variations in amino acid concentrations have been associated with aphid resistance in peas (Auclair et al., 1957), wheat (Kazemi & van Emden, 1992), and barley (Weibull, 1987).Schmidt et al. (2007) suggested that alfalfa living mulch reduces plant quality for aphids on soybean, associated with a reduction in nitrogen concentration in soybean leaves.
In this study, we aimed to assess potential impacts of rolled-rye cover crop on N. ribisnigri, its natural enemies, and alternative prey (other herbivorous species attacked by natural enemies associated with aphids) in lettuce production.In situ, we assessed the impact of rolled-rye cover crop on field colonization and abundance of the insect fauna via non-destructive scouting and yellow pan traps in two lettuce crops.We also used exclusion cages to investigate potential effects of rolled-rye cover crop on lettuce quality (biomass and amino acid content) and N. ribisnigri fitness parameters (development, adult size, and abundance).Under greenhouse conditions, we examined potential lingering effects of rye on lettuce and N. ribisnigri by growing lettuce on field soil sampled underneath the cover crop.

Field site
Field work was conducted in summer 2021 at the AAFC experimental farm in Ste-Clotilde-de-Châteauguay on Histosol.Experiments were first conducted on head lettuce (cv.Estival) transplanted on 9 June and harvested on 23 July, and next repeated on romaine lettuce (cv.Sunbelt) transplanted on 2 August and harvested on 21 September.Experimental design consisted of two treatments, a cover crop composed of rolled-rye and a control (bare ground) replicated 4× each for a total of eight lettuce plots (11 × 11 m).Plots were subdivided in six beds (1.8 m wide) containing four rows of lettuce each.Rows within a bed were spaced 40 cm apart, and each lettuce within a row was spaced 30 cm apart (total of 80 lettuce plants per plot).Barley (Hordeum vulgare L., Poaceae) was sown between the plots, spaced 10 m apart, and regularly mowed to maintain a short vegetative cover.The control treatment was a conventional lettuce production tilled in bare soil.Irrigation and fertilization followed local lettuce production practices (Parent & Gagné, 2010).In the cover crop treatment, rye (S. cereale cv.Gauthier) was sown on 27 September 2020 at a seeding rate of 230 kg ha −1 .The rye was roller-crimped twice when anthesis was completed (stage 69 on the Zadoks growth scale) (Zadoks et al., 1974), which occurred 3-4 days before the first lettuce transplant (8 June).Roller crimping failed to completely terminate early season rye, which remained green and bushy for the first lettuce crop experiment (cv.Estival, hereafter 'early rolled-rye experiment' or 'early').The glyphosate-based herbicide Roundup was applied on 27 July (7 days prior to lettuce transplant) to achieve complete termination of rye cover crop.In the second lettuce crop experiment (cv.Sunbelt, hereafter 'late rolled-rye experiment' or 'late'), the fully terminated rye had a significant reduction in thickness due to decomposition, but still covered ca.90% of the ground (Figure S1).All plots were manually weeded weekly, on the same day.Fungicides were applied 3× during the growing season to control mildew [mandipropamid (Revus) on 25 June, fluopicolide (Presidio) and fosetyl-Al (Aliette) on 27 August] and prevent lettuce losses.No insecticides were applied during the growing season.Soil temperature and humidity were recorded at surface level and 15 cm below ground using EM50 data loggers with 5TM, 5TE, and GS3 probes (METER Group, Pullman, WA, USA) at the center of each plot in the early rolled-rye (Figure S2) and late rolled-rye (Figure S3) experiments.

Plant colonization by aphids and their natural enemies
Non-destructive visual observation and yellow pan trapping were used to assess the presence of aphids and their natural enemies.For visual observation, twice a week, from 16 June to 13 July (early; nine scouting dates) and from 10 August to 2 September (late; seven scouting dates), 30 randomly selected lettuce plants were observed (30 lettuce plants × 4 plots × 2 treatments).For each lettuce, the entire plant was meticulously inspected by gently flipping over each leaf to count all the insects.Scouting was conducted until one of the two following conditions were met: (1) ⅔ of the sampled lettuce plants were infested by lettuce aphids, or (2) heads of lettuce were completely formed (phenological development stage BBCH = 45) (Jenni & Bourgeois, 2008).Aphids typically associated with lettuces were morphologically identified to the species level based on Capinera (2001), whereas adult and larval natural enemies were pooled into higher taxa.As the presence of a single aphid per plant can downgrade the quality of lettuce, plant colonization data were expressed in terms of the percentage of plants with N. ribisnigri or aphid natural enemies at the scale of the experimental plot.
Yellow pan traps (27 cm diameter) filled with soapy water (1:100) were deployed in the center of each plot at ground level, every week for 24 h.All adult and larval insects captured in the traps were sorted and preserved in ethanol (95%) for further identification.Lettuce aphids, their natural enemies, and alternative prey (other aphid species not associated to lettuce, thrips, and psyllids) were morphologically identified to the species level when possible, otherwise family or order levels were used (Neuroptera : Froeschner, 1947;Anthocoridae: Kelton, 1978;Coccinellidae: Gordon, 1985;Thysanoptera: Mound & Kibby, 1998;Capinera, 2001).The total number of insects per trap and per date was calculated for three groups: (1) N. ribisnigri, (2) aphid natural enemies, and (3) alternative prey.
To increase the density of lettuce aphids in the surrounding environment, a total of 700 N. ribisnigri winged individuals from the rearing colony were released.The releases were carried out at 20 introduction points spaced 20 m apart, located around the periphery of the experimental field and within 5 m of the outermost plots.Aphids were released on 22 June (early) and on 16 August (late).To accomplish this, 35 winged individuals were placed in 50 mL Falcon plastic tubes planted in the soil at each releasing point.The lids were next unscrewed to allow the aphids to take flight.After 30 min, all aphids had taken flight.

Exclusion cage experiment
The impacts of rye cover crop on N. ribisnigri (population growth, aphid size, and apterous/alate morph) and lettuce quality (biomass and amino acid content) were measured under field conditions using exclusion cages.Approximately 1 week after transplanting the lettuce, on 17 June and 13 August, 80 plants at the 6-9 leaf stage were randomly selected in the experimental design described above (10 replicates × 4 plots × 2 treatments: control and rolled-rye cover crop).Each lettuce was carefully observed and swept with a fine brush to remove any arthropods prior to aphid inoculation.Five N. ribisnigri (third to fourth instars) from the laboratory colony were inoculated in the center of each lettuce and then covered with an exclusion cage to prevent the entry of predators or parasitoids.Exclusion cages consisted of a small tent (15 × 15 × 30 cm) made of muslin fabric (mesh size 0.2 × 0.2 mm) supported by two U-shaped metal rods firmly anchored into the soil.Ambient temperature and temperature within the cages were recorded throughout the experiment (every 15 min) using HOBO UA-002-64 temp/light loggers (one logger for ambient temperature, one logger for exclusion cage over the control, and one logger for exclusion cage over the rolled-rye cover crop) (Onset, Bourne, MA, USA) (Figure S4).
After 11 days, a period allowing the inoculated aphids to reach the adult stage and produce a first generation of mature offspring (Diaz & Fereres, 2005), exclusion cages were removed and each lettuce was carefully collected in a plastic bag and brought back to the laboratory.To determine the number of aphids per plant, each leaf was detached and examined for the presence of N. ribisnigri.All aphids were counted and classified according to their developmental stage (juvenile/adult) or their morph (winged/apterous).The total number of aphids and the proportion of winged aphids were compared between the two treatments for both experiments.All apterous adults were preserved in ethanol (95%) in order to measure tibial lengths as a fitness proxy and fresh plant samples were used to estimate lettuce quality (see below for methodologies).

Greenhouse experiments
Greenhouse experiments aimed to examine the impact of rolled-rye cover crop on soil microbiota, as well as any potential effects on lettuce quality and aphid fitness.To this end, on 7 September 2020, we collected 60 L of the first 20 cm of soil from rolled-rye plots (aboveground biomass without the cover crop) and 120 L of soil from the control plots (bare ground) from the field site.Soil samples were sieved (1 × 1 cm mesh size) to remove debris and macro-arthropods.They were then mixed (1:1) with general purpose PRO-MIX BX (Premier Tech Horticulture, Rivière-du-Loup, QC, Canada) without mycorrhizae to minimize nutrient heterogeneity (Kos et al., 2015).To assess more precisely the magnitude of an eventual effect of the rolled-rye cover crop on soil microbiota and its repercussions on lettuce plants and aphids, a third soil treatment was established by pasteurizing (20 min at 70 °C) 60 L of the control soil mix to reduce soil microbiota biomass (Weller et al., 2002).The three soil treatments (rolled-rye cover crop, control, and pasteurized control) were then divided in 135 1.5-L pots (45 replicates × 3 soil treatments) where lettuce plants at the 2-4 leaf stage were transplanted.Lettuce plants were kept at 18 °C on a L16:D8 cycle under a weekly fertilization program: 115 mg of 6-11-31 (N-P-K) fertilizer, 26 mg magnesium sulfate at 9.8%, 85 mg calcium nitrate, and 40 mg nitrogen (total 100 mL per week).
Seven days after lettuce transplanting, five third to fourth instar N. ribisnigri were inoculated on 45 plants (15 replicates × 3 treatments).As aphid feeding is known to alter the quality of the host plant (Sandström et al., 2000), aphid presence/absence was added as a factor to the experimental design.A cage made of a plastic jar with large ventilation holes covered with muslin (0.2 × 0.2 mm mesh size) was placed on each lettuce to prevent aphid movement between plants.Plants without aphids were also covered with a cage to provide similar growth conditions.After 7 days, cages were removed and the number of aphids (juvenile/adult) and their morph (winged/apterous) were recorded as described above for the exclusion cage experiment.
To assess the impact of soil treatments on developmental rate of N. ribisnigri, we measured the intergenerational time (number of days from birth to offspring production) in a second experiment.To this end, 45 mature apterous aphids were collected on lettuce grown in each soil treatment and reintroduced on new lettuce plants from each treatment.Each individual aphid was placed in a ventilated clip-cage sealed with petroleum jelly (Vaseline; Unilever, London, UK) (3 clip-cages per plant × 15 lettuce plants × 3 soil treatments).Inoculated aphids were left 72 h in clip-cages to ensure production of offspring.Following adult removal, juvenile aphid population within each cage was recorded (number of aphid and developmental stage) twice a day (08:00 and 15:00 h) until the recording of the first new generation of juveniles (i.e., inter-generational time).At the end of experiment, all apterous adults were preserved in ethanol (95%) for tibial length measurements.

Tibial length measurements
Beside developmental time and aphid abundance, tibial length was used as a size proxy for N. ribisnigri (Lanteigne et al., 2014).In both field and greenhouse experiments, the two hind legs of adult apterous aphids were severed at the base of the femora and then mounted on a slide.Apterous aphids were selected over winged aphids because they had completed their entire development under a given treatment and to avoid size variations between morphs (alate vs. apterous).Measurements were performed using a VHX-S660E (Keyence, Osaka, Japan) free-angle observation system at 100× magnification from the top of the femora-tibia articulation to the insertion point of the first tarsomere.Tibial length of both hind legs was then averaged per individual.

Lettuce quality
To assess nutritional indices of lettuce and its quality for N. ribisnigri, leaf amino acid profiles and dry biomass were measured from lettuce grown under field and greenhouse conditions.Concentrations of amino acids pertinent to aphid nutrition [arginine (arg), histidine (his), isoleucine (ile), leucine (leu), lysine (lys), methionine (met), phenylalanine (phe), threonine (thr), tryptophan (trp), tyrosine (tyr), and valine (val)] and other non-essential amino acids [alanine (ala), asparagine (asn), aspartic acid (asp), glutamic acid (glu), glutamine (gln), proline (pro), serine (ser)] were used as a proxy bridging lettuce nitrogen content and nutritional quality (Sandström et al., 2000).Lettuce metabolomes were characterized from two leaf disks (8 mm diameter) punctured from the third youngest leaf and then preserved at −80 °C.In field experiments, four randomly selected lettuce plants per plot were sampled (2 disks × 4 lettuce plants × 4 plots × 2 treatments × 2 experiments), whereas in greenhouse experiments, eight randomly selected lettuce plants per treatment were used (2 disks × 8 lettuce plants × 3 soil treatments).To extract the metabolome, leaf material was crushed using a bead beater and the resulting mixture underwent centrifugation to separate water-soluble metabolites and remove debris.The McGill Goodman Cancer Research Center (Montreal, QC, Canada) conducted further analyses to quantify amino acid concentrations, following method B from Fu et al. (2021).The extracted samples were analyzed using ultra performance liquid chromatography/triple quadrupole mass spectrometry (UPLC-QQQ-MS/MS) (Agilent Technologies, Santa Clara, CA, USA), with 5-μL injections, utilizing a 3μm Intrada-AA column (3.0 × 150 mm) (Imtakt, Kyoto, Japan) with a unison guard for chromatographic separation.Following leaf disk extractions, lettuce biomass per plant was determined as follows: aerial parts of the plants (leaves and heart) were dried in a forced air oven for 96 h at 60 °C, and dry weight was next measured using an analytical balance (OHAUS precision standard; Ohaus, Florham Park, NJ, USA).

Statistical analyses
We performed all statistical analyses in R v.4.0.4 (R Core Team, 2021).For field experiments, as the two lettuce cultivars were grown on different dates, they were analyzed separately.After visually inspecting data repartition, a modeling family was chosen.For Gaussian family, assumptions of homoscedasticity were tested by inspecting residual vs. fixed plot along with Levene's and Bartlett's homogeneity of variance tests (leveneTest and bartlett.test in car and stats package) whereas normality of residuals was assessed both visually and using Shapiro-Wilk's normality test (function shapiro.test,stats package).For the number of colonized lettuce plants, we used a generalized linear model (GLM) following a Poisson distribution.In this model, 'treatment' and 'date' were designated as fixed effects, whereas 'plot' was included as a random effect with a nesting structure within the 'date' variable.Overdispersion and zero-inflation were visually assessed and tested using the DHARMa package (functions testDispersion and testZeroInflation) on calculated scaled residuals (function SimulateResiduals) (Hartig, 2022).The quasi-Poisson distribution was used to correct for overdispersion, only for lettuce plants colonized by natural enemies in the early rolled-rye experiment.For yellow pan trap data, we specified GLMs following a similar structure as the one employed for plant colonization.Numbers of N. ribisnigri trapped were too small (<3 aphids per trap) to allow for statistical analysis for both experiments.Numbers of natural enemies and alternative prey per trap were modeled using a zero-inflated GLM following a negative binomial distribution in the early rolled-rye experiment (using the package glmmTMB), and a Poisson distribution in the late rolled-rye experiment.Fixed effects significance was assessed by analysis of deviance table (type II test with function ANOVA from car package).When a significant interaction between treatment and date was observed, treatment effect was tested again separately for each date with the same model.
For aphid population growth in exclusion cages, we used a non-parametric Kruskal-Wallis test for both early and late rolled-rye experiments.Lettuce plants that had been contaminated by other aphid species (0 in the early and 22 in the late rolled-rye experiment) were excluded from the analysis because other aphids could have competed with N. ribisnigri.Tibial length was analyzed with a type II ANOVA in the early rolled-rye experiment (148 pairs in control and 42 pairs in rolled-rye cover crop), whereas in the late rolled-rye experiment a non-parametric Kruskal-Wallis test was used (49 pairs in control and 26 pairs in rolled-rye cover crop).Due to high variability in tibia length, which could be attributed to measurement errors, we utilized the 'outline' argument within the boxplot() function to remove outliers.This enabled us to remove 22 outliers in the early rolled-rye experiment, thereby achieving the necessary normality required for conducting the ANOVA test.No outliers were observed in the late rolled-rye experiment.
For modelling aphid population growth in the greenhouse, we used linear mixed models (LMMs), with the 'treatment' variable as a fixed effect, and 'position' integrated as a random effect, utilizing a specified grouping structure.Mean tibial lengths were pooled per treatment and treatment effect was assessed using a linear model with cube transformed lengths.In the inter-generational time experiment a non-parametric Kruskal-Wallis test was used, with the 'outline' argument within the boxplot() function to remove outliers (nine data points removed).The same analysis was used to test the number of juvenile aphids in the clip-cages at the start of the experiment and the proportion of these aphids that remained alive at the onset of offspring production.
Amino acids molarity (M) from both field and greenhouse lettuce leaf disks was visually compared between treatments using nonmetric dimensional scaling (NMDS) with the metaMDS function from the vegan package.Amino acid concentrations based on Bray-Curtis dissimilarity matrix were then compared using permutational multivariate analysis of variance (PERMANOVA; function adonis, vegan package).Assumption of within group homogeneity of dispersion was verified using ANOVA (function anova, stats package) on multivariate homogeneity of group dispersions (function betadisper, vegan package).Following significant PERMANOVA in the early rolled-rye experiment, each amino acid concentration was tested separately with a type II ANOVA (function anova).All amino acid concentrations were log transformed except for Thr, Val, Asn, Ala, Arg (square root transformed data), and Lys, which were compared with a non-parametric Kruskal-Wallis test.In the greenhouse experiment, as aphid infestation effect was not significant, type I ANOVAs on log transformed concentrations were used instead, followed by post-hoc pairwise comparisons using the Tukey Honest Significance Differences (HSD) method (function TukeyHSD, stats package).Lettuce dry weights between treatments were compared using Kruskal-Wallis tests and type II Wald χ 2 test for field and greenhouse experiments, respectively.

Yellow pan trapping
In the early rolled-rye experiment, there was a significant interaction between treatment and date on the abundance of aphid natural enemies (interaction: χ 2 = 29.00,d.f.= 6, P < 0.001).On most sampling dates, more natural enemies were captured in rolled-rye cover crop plots than in control plots (Figure 2A).On 6 July, we observed an unusual high number of adult syrphids in control plots (109.5 ± 30.1) (mean ± SE), accounting for 67.3% of all syrphids captured in this treatment over the experiment.Overall, traps in rolled-rye cover crop plots captured almost 8× more parasitoids (Aphelinidae and Aphidiinae, 167 vs. 22 individuals) and 6× more anthocorids (Orius spp., 75 vs. 12 individuals) than traps in control plots (Figure 2B).
The abundance of alternative prey in the early rolled-rye experiment was influenced by treatment and date interaction (interaction: χ 2 = 12.98, d.f.= 6, P = 0.043).More alternative prey were captured in rolled-rye cover crop plots than in control plots, except on 22 June (Figure 4A).Numbers of thrips and aphid species not typically associated with lettuce plants ('other aphids') greatly increased over time in the rolled-rye cover crop treatment compared to the control treatment, with almost 4× more thrips (1485 vs. 392) and 3× more 'other aphids' (377 vs. 123) (Figure 4B).
On lettuce from the late rolled-rye experiment, aphid abundance was also lower in the rolled-rye treatment F I G U R E 3 (A) Mean (± SE) number of aphid natural enemies per yellow pan trap throughout the season in 2021 in control (red) and rolled-rye cover crop (blue) plots for lettuce from the late rolled-rye experiment and (B) total abundance per taxon.There were no significant differences.compared to the control (8.7 1.5 vs. 20.7 ± 3.2 aphids; K-W test: χ 2 = 13.21,d.f.= 1, P < 0.001).There was no significant difference in the proportion of alates between rolled-rye cover crop and control treatments (K-W test: χ 2 = 1.19, d.f.= 1, P = 0.28).The tibias of adult apterous aphids were also shorter under the rolled-rye treatment compared to the control (1526.0 ± 15.8 vs. 1595.6± 15.8 μm; K-W test: χ 2 = 6.76, d.f.= 1, P = 0.009).

Lettuce quality
Under field conditions, at the end of the early rolled-rye experiment, lettuce plants growing in the rolled-rye cover crop plots were almost 3× smaller than lettuce plants from control plots (1.31 ± 0.1 vs. 3.76 ± 0.09 g; χ 2 = 55.85,d.f.= 1, P < 0.001).The rolled-rye cover crop treatment resulted in lower leaf concentrations of all amino acids tested in comparison with the control treatment, and the profiles of these amino acids was significantly different (F 1,25 = 65.15,R 2 = 0.764, P < 0.001; Figure 6A).

DISCUSSION
Our study provides evidence that rolled-rye cover crop applied in cultivated Histosols has significant effects along the trophic chain, influencing interactions between lettuce, herbivores, and their natural enemies.Rolled-rye cover crop decreased lettuce colonization by winged N. ribisnigri and subsequent aphid population growth in two successive lettuce crops.On the other hand, in the first experiment (early rolled-rye experiment), cover crop plots recruited more aphid natural enemies and alternative prey than control plots, likely delaying the spread and growth of N. ribisnigri populations.Furthermore, aphid fitness was reduced for individuals developing on lettuce plants from the rolled-rye cover crop treatment, suggesting a decrease in lettuce quality for N. ribisnigri.Competition between lettuce and the incompletely terminated rye early in the season reduced growth and nutritional quality of the first lettuce crop (cv.Estival).Such competition poses agronomic challenges regarding rye termination and lettuce production.Controlled condition experiments aiming to detect lingering effects of rolled-rye cover crop within soil did not reveal any effects on abundance and fitness of N. ribisnigri nor on lettuce growth, indicating that persistent effects were too subtle to be detected, of short duration, or influenced by other mechanisms.

Impacts on aphid population dynamics
Live early rolled-rye cover crop took a green bushy aspect whereas the fully terminated late rolled-rye was straw like.Despite these striking differences between the two rolledrye cover crop experiments, they both reduced N. ribisnigri colonization compared to their respective control.In the early rolled-rye cover crop, lettuce plants were colonized by up to 22% less N. ribisnigri, similar to the 24% reduction of A. glycines in soybeans covered by rolled-rye cover crop observed across Minnesota, USA (Koch et al. 2012).For letplants grown on late rolled-rye with higher aphid densities, a reduction of 49.2% N. ribisnigri was observed.Such a pattern might result from structural changes within the crop that cause visual interference during host plant location by N. ribisnigri and natural enemy recruitment.
Host plant visual interference may have played a role in reducing lettuce colonization by N. ribisnigri in both cultivars.Positive phototaxis contributes to alate aphid orientation with several species being attracted to light in the ultraviolet (UV; 320-330 nm) and green-yellow (550-590 nm) spectrum (Kirchner et al., 2005;Döring & Chittka, 2007).Contrast between the plant and its background can therefore be critical to host finding by herbivores.Artificial reflective mulches may interfere with plant location in insects, including aphids in different crops (Zalom, 1981;Brown et al., 1993).Comparing a living buckwheat (Fagopyrum esculentum Moench) cover crop to a polyethylene mulch, Nyoike & Liburd (2010) demonstrated that only the living cover crop reduced colonization by aphids and whiteflies on zucchini squash and increased natural enemy recruitment.In our lettuce plots, the lack of difference during late rolled-rye experiment in insect colonization between the two treatments could result from the loss of contrast between cover crop and soil following the rolled-rye decomposition, resulting in lower reflectance at the end of the growing season (Wagner-Riddle et al., 1996).
Along with reduced aphid colonization in the early rolled-rye experiment, we observed an increase up to 35.9% in foliar natural enemy colonization.Yellow pan traps also collected 26.6% more natural enemies in rolledrye cover crop plots, including generalist predators and specialized aphid parasitoids.Yellow pan traps also revealed a higher abundance (up to 73.6%) of alternative prey, mainly thrips and other aphid species, that likely attracted and retained generalist predators.Top-down pest regulation service from plant resource diversification, the enemies hypothesis (Root, 1973), has been described in other crop-cover crop associations.Schmidt et al. (2004) reported that the addition of a barley cover crop lowered densities of several cereal aphid species on wheat by up to 25%, correlating with an enhanced presence of natural enemies such as aphid midges, hoverflies, and spiders.Similarly, Kahl et al. (2019) reported that a living red clover cover crop recruited an assemblage of predatory bugs and coccinellids, reducing densities of the melon aphid, Aphis gossypii Glover, on cucumbers.
In the late rolled-rye experiment, however, neither visual scouting nor trapping detected important changes in the abundance of natural enemies or alternative prey despite higher aphid abundance.This situation could result from rye cover crop being much more degraded than earlier in the season, thus reducing the attraction of foliar natural enemies.However, we cannot exclude the potential impact of ground beetles (not sampled in our study), which populations could have been favored by the presence of rye straw on the ground.Koch et al. (2012Koch et al. ( , 2015) ) reported suppression of A. glycines populations on soybean in rye cover crop plots, with no correlation with Orius insidiosus (Say) and coccinellid abundance, suggesting that other host plant-related effects might be at play.Rivers et al. (2020) observed that a cover crop mixture of hairy vetch (Vicia villosa Roth), triticale (× Triticosecale spp.) and rye (S. cereale) in corn significantly increased the density of ground dwelling predators such as spiders and carabid beetles.Also in corn field covered with a rye cover crop, Hartwig & Ammon (2002) reported higher abundance of ground dwelling predators together with a reduction of black bean aphid, Aphis fabae Scopoli, densities.

Impacts on Nasonovia ribisnigri fitness
Early in the season, live rolled-rye cover crop reduced N. ribisnigri fitness on lettuce grown in field cages, with adults having 3.5% shorter tibias and populations carrying 13.2% more alates than in the control treatment.For the late rolled-rye cover crop, tibial lengths were only reduced by 4.4% in aphids that developed on lettuce with rolled-rye compared to the control.Although statistically significant, the decrease in tibial length is minor compared to the 10% reduction associated with partial N. ribisnigri resistance in lettuce plants observed by Lanteigne et al. (2014).Nevertheless, increased alate production also indicates conditions less favorable for aphid development and reproduction.Such observed reductions in N. ribisnigri fitness are more likely to be attributed to the lower nutritional quality of lettuces grown in the rolled-rye cover crop treatment.Other common factors that can cause stress to aphids, such as overcrowding, thermal stress, and the presence of natural enemies in the habitat, may have also played a role (Mittler & Sutherland, 1969;van Emden & Harrington, 2017;Sentis et al. 2017).

Impacts on lettuce quality
We relied on amino acid profiles and lettuce dry weight to detect rolled-rye cover crop mediated changes in lettuce quality and the consequences on N. ribisnigri fitness.Early season live rolled-rye cover crop drastically reduced both lettuce dry weight (65.2%) and leaf concentration of all tested amino acids.A reduction of amino acid concentrations points towards nitrogen stress as nitrogen would be rapidly depleted when producing nutrient uptake protein (Atilio & Causin, 1996;Broadley et al., 2000;Ciriello et al., 2021).In contrast, stress unrelated to nitrogen deficiency, such as osmotic and oxidative stress caused by drought, leads to the accumulation of amino acids associated with osmotic regulation like pro, whereas phosphorus or potassium deficiency are usually associated with the accumulation of nitrogen rich compounds (Evans & Sorger, 1966;Atilio & Causin, 1996;Tausz et al., 2004;Criado et al., 2017).Late season terminated rolled-rye cover crop had a smaller negative impact on lettuce dry weight (36.5%) and did not affect acid profiles.Although the negative consequences of cover crop on plant suitability for aphids have rarely been formally tested, they are assumed to occur.In soybean growing on living alfalfa (Medicago sativa L.) cover crops, Schmidt et al. (2007) reported both a natural enemy dependent delay in A. glycines establishment and a bottom-up suppression of aphid population growth in exclusion cages.They correlated the latter with lower nitrogen concentration in soybean leaves, which was attributed to competition between the cover crop and the crop.The high carbon-to-nitrogen ratio of rye litter could also lead to nitrogen immobilization, although nitrogen availability in southern Québec Histosols may be sufficient to meet lettuce requirements (Dessureault-Rompré et al., 2020).Other factors such as competition for light in early living cover crop, water availability, and rye born allelopathic compounds may also have played a role in determining lettuce overall quality.
Cover crop affects soil microclimate, microbiota, and nutrient dynamics (Higo et al., 2019;Kim et al., 2020).By collecting soil from underneath rolled-rye cover crop and control plots at the end of the lettuce growing season, we aimed under controlled conditions to detect any lingering effects on soil microbiota that might affect lettuce quality and N. ribisnigri fitness.Leaf amino acid profiles were only affected by pasteurization, confirming that soil microbiota modify the lettuce leaf metabolome.Kabouw et al. (2011) and Bezemer et al. (2005) have also linked soil microbiota with plant-aphid interactions.In our study, rolled-rye cover crop did not have detectable impacts on soil properties and N. ribisnigri fitness at the end of the growing season.
Both living and terminated rolled-rye cover crops contribute to reducing populations of N. ribisnigri on lettuces.However, of significance, these positive effects did not extend to lettuce production.In fact, the presence of living rolled-rye cover crop early in the season strongly inhibited lettuce growth and quality and further research is needed to identify the optimal methodology for minimizing the negative impact of the rolled-rye cover crop on crop production.Despite this limitation, rolled-rye cover crop still holds potential as a promising agronomical practice for mitigating erosion of Histosols and controlling aphid pests.However, it's important to underline that our field study was conducted in only one location over the course of one year, covering two consecutive lettuce production cycles.In order to develop a more comprehensive understanding of the pest-suppressive effects of the rolled-rye cover crop, it is imperative to conduct additional research across various locations and over multiple years.

F
I G U R E 2 (A) Mean (± SE) number of aphid natural enemies per yellow pan trap throughout the season in 2021 in control (red) and rolledrye cover crop (blue) plots for lettuce from the early rolled-rye experiment and (B) total abundance per taxon.The asterisks indicate significant differences between treatments (Wald χ 2 test: * 0.01 < P < 0.05, ** 0.001 < P < 0.01, *** P < 0.001).F I U R E 4 (A) Mean (± SE) number of alternative prey per yellow pan trap throughout the season in 2021 in control (red) and rolled-rye cover crop (blue) plots for the early rolled-rye experiment and (B) total abundance per taxon.The asterisks indicate significant differences between treatments (Wald χ 2 test: ** 0.001 < P < 0.01, *** P < 0.001).

F
I G U R E 5 (A) Mean (± SE) number of alternative prey per yellow pan trap throughout the season in 2021 in control (red) and rolled-rye cover crop (blue) plots for the late rolled-rye experiment and (B) total abundance per taxon.The asterisks indicate significant differences between treatments (Wald χ 2 test: *** P < 0.001).