Experiments were conducted during the crop growing (vegetative) seasons of 2015 and the 2016 in Central and Eastern Transylvania, Romania in order to assess the population density of the most important invasive weed species and infesting virus-vectoring aphids, both from low- and high-input agricultural crops.

Low-input, traditionally managed fields (LIF). This area belongs to a traditionally managed field (low-input) of the Saxon cultural region encompassing an area of 7,440 km² at altitudes between 230 and 800 mm above sea level (a.s.l.) and characterized by a landscape mosaic of different land-cover types (28% forest, 24% pasture, and 37% arable land, mostly maize, potato and alfalfa). The farming practices in the studied area were predominantly small scale, with no chemical inputs and for subsistence purposes. One consequence of this land use is the exceptional biodiversity and natural value of the farming landscape (Akeroyd and Page 2001). However, the being not economically viable, the abandonment of croplands in this region is frequent, this resulting in the establishment and high abundances of invasive weeds (Zimmermann et al. 2015).

High-input, conventionally managed fields (HIF). This study region contrasts the previously described region by large monocultures and farming landscapes with low levels of natural vegetation and heterogeneity (Eastern Transylvania). The area of about 5,500 km² was intensively treated with synthetic fertilizers and pesticides, major crops were maize, potato and alfalfa (Table 1).

The studied fields from the two regions were situated in the same altitudinal range of about 250 m a.s.l. and under comparable bioclimatic conditions. The distance between the studied areas was about 200 km.

Three weed and two native aphid species were studied, these being the most common species in the study area. The most important weed species, all of the family Asteraceae, were the annual fleabane, Stenactis (=Erigeron) annua (L.), Canadian horseweed, Erigeron (=Conyza) canadensis (L.), and the goldenrod, Solidago canadensis (L.). These species are known to use a diverse range of habitats and are considered economically important weeds in Europe (Anastasiu and Negrean 2005). Stenactis annua is often a dominant species within the invasive weed communities, and has been reported from almost all European countries, its expansion showing an increase (Edwards et al. 2006; Tunaitienė et al. 2015; Pacanoski 2017). Erigeon canadensis is an annual plant native throughout most of North and Central America. It is also widely naturalized in Eurasia (Nandula et al. 2006; Shah et al. 2014; Bajwa et al. 2016). Solidago canadensis is a perennial weed native to north-eastern and north-central America, but has established as an invasive plant throughout Europe (Abhilasha et al. 2008; Fenesi et al. 2015).

The two native aphid species included in this study where the highly polyphagous leaf-curling plum aphid, Brachycaudus helichrysi (Kaltenbach) and the similarly polyphagous foxglove aphid, Aulacorthum solani (Kaltenbach). These are particularly important species, not only because of their wide host plant range but also for their diverse virus transmission. The host plant range of B. helichrysi includes members of the Asteraceae, e.g. Chrysanthemum, species of Prunus and also species of Solanum, Fragaria, Trifolium, Medicago, and Citrus and maize (Tatchell et al. 1983; Powell et al. 1992; Isac et al. 1998; Popkin et al. 2017). Viruses transmitted by these aphids include plum pox, Potato virus Y and the Beet mild yellowing virus (Isac et al. 1998). Host plants of A. solani includes tomato, peppers, tobacco, celery, carrots, tulip bulbs, cucurbits and legumes (Tatchell et al. 1983; Jandricic et al. 2014). Of transmitted viruses, the most important are Potato viruses A, Y and X and Potato leaf roll virus, Cucumber mosaic virus, Soybean dwarf virus, Bean yellow mosaic virus and Turnip yellows virus (Jandricic et al. 2010, 2014).

First, we selected two study sites in each region, these being 10 km in a fist and 15 km distant in a second region from each other. In each site we established two transects (at least at 1 km apart) of 10 m long × 1 m wide at an approximately equal distance of at least three major crops (maize, potato and alfalfa). In this way, each transect was surrounded by at least 8–10 ha of high-input, and at least 0.5–3 ha area of low-input, agricultural crops. Each transect was carefully measured and located using GPS. Second, inside each transect we placed ten 1 m² quadrats. Each of these quadrats was further subdivided in 10 × 10 cm plots (100 for each quadrat) and all plants (native and invasive) were counted and their coverage estimated within them. (Andújar et al. 2010). Third, ten invasive weed plant individuals from each quadrat were randomly collected in plastic bags. The number of invasive plants collected for each species from each quadrat mirrored the coverage of the species within the quadrat. We subjectively decided that we will collect at least eight plants when the coverage of a given species in a quadrat was at least 80% and up to two plants if the coverage of the species was up to 20%. We decided upon these percent coverage thresholds because in each quadrat there was one highly dominant invasive plant species (its coverage having at least 80%) and one species which had a coverage between 15–20%. Therefore from each quadrat, out of the 10 plants at least eight belonged to the dominant species and 1 or 2 to the second dominant species.

Because plants contained aphid colonies, and the exact number of individual aphids was important, all bags were labelled and kept at low temperature (near 0–4 °C in a cool box), then returned to the laboratory, whereupon all samples were stored at −20 °C, and aphids counted and species identified (Blackman and Eastop 2000; Blackman 2010). In total, 100 plant samples were collected per transect and management system (400 samples per management system per collection data). Assessment started at the end of May and repeated fortnightly five times during the summer until the end of the weed vegetative season, whereupon no more aphids were found. The whole procedure was repeated the following year using the same collection methods within the same transects. All aphids were carefully counted under laboratory conditions, and the various species identified).

Leaf samples used for enzyme analyses were collected each year from each abundant weed species per experimental field, sub-area and transect, starting from the first until the last assessment. Separate young leaves, all containing aphids, were collected from the weeds (n = 10 samples / 1 m² sub-transect = 100 / transect). Samples were also held at −20 °C until enzyme extraction and activity assays.

For extraction, 400 mg of frozen leaves were homogenized in 1 ml of 50 mM phosphate buffer, pH 7.0, using a FastPrep Instrument high-speed benchtop homogenizer (MP Biomedicals). The homogenate was centrifuged at 6,500 r.p.m. for 20 minutes at 4 °C, and the supernatant collected. Protein concentration of the enzyme extract was determined by the Bradford method (Bradford 1976). POD activity was determined by the method of Németh et al. (Németh et al. 2009). The reaction mixture contained 955 µl of 50 mM phosphate buffer, pH 5.5, 10 μl of 0.01 g/l 3.3'-diaminobenzidine and 30 µl of enzyme extract. The reaction was initiated by the addition of 5 μl 0.3% hydrogen peroxide. The increase in absorbance at a wavelength of 480 nm was followed in a spectrophotometer for 5 minutes and 5 and 10 μmol min⁻¹ · mg protein^-1 unit of POD activity was separately defined as an absorbance change of 0.01 units·min ^-1.

The experiment was performed during the vegetative period of 2017 by setting-up 30 blocks of the two most abundant weed species, E. canadensis and S. annua and the most frequent crop plants cultivated (maize, potato and alfalfa; Fig. 1). Because of a relatively low density of plants and high aphid density variation on S. canadensis, this weed was not included in the colonization experiment. Specimens of E. canadensis and S. annua of the same age (maturity) were collected in April from the field and potted in 8 litre pots.

Figure 1.

Aphid colonization experiment design, weed plants of E. canadensis or S. annua surrounded by two maize, two potato and two alfalfa plants.

After acclimatization in May, similar size plants of about 30 cm were selected for experiments. Crop plants of maize, potato and alfalfa were also cultivated in 8 litre pots, and similar sized plants selected in May for experiments. All weed and crop plants were first cleared of any infesting aphids by visual checking of all leaves and shoots. In the case of any aphid colony being detected, these were removed by brushing off colonies from the plants with a soft paint brush. If other insect species were detected, these were also removed. Insect-cleared plants were then allocated for experiments. Altogether 30 experimental blocks were set-up, 15 blocks with E. canadensis and 15 with S. annua under open field conditions where no other weeds and similar crop plants were present. Weed plants were placed at a distance of 20 m from each other, and six crop plant (two maize, two potato and two alfalfa) were placed around one weed plant to a distance of 50 cm (Fig. 1). Weed plants from each blocks were then artificially colonized with aphids by collecting B. helichrysi and A. solani from naturally occurring E. canadensis and S. annua plants. Leaves or shoots of weeds with aphid colonies were carefully removed, the colony of each aphid species reared to five individuals of 4^th-instar nymphs by carefully brushing down all other individuals. Thereafter, the plant leaf or shoot was placed to the top of the experimental weed plant already placed in blocks. Each weed plants received two colonies (five aphids of each) from both aphid species.

The established aphid colonies (assessed by careful visual assessment over a 10 minute period as to whether aphids were feeding consistently on plants and not moving) were checked after 24, 48 and 72 hours. If no colony establishments were detected, new colonies were placed on the weeds. The aphids were then left to reproduce for 10 days. The assessment of aphids started after 14 days after aphid colony establishment, such that enough winged individuals were present to leave weeds and colonize crop plants. Aphid numbers were assessed on both weed and crop plants of the same blocks starting from mid-May as follows: two randomly selected blocks (one with E. canadensis and one with S. annua plants) were sampled by enclosing the infested plant in a transparent polythene bag and then cutting this free with scissors or a knife.

On return to the laboratory, the entire content was stored at −20 °C and the next day all samples were carefully assessed for aphids and their respective numbers counted under a stereo microscope. By this means, all individual aphids were captured and counted. The same procedure was repeated the following day, until the total number of blocks and plants were sampled by cutting all and aphids from plants counted in lab. The entire sampling was done within a two-week period and completed by the end of May, a time when aphid migration to new host plants occurs. All aphids were counted, recorded regarding the weed and crop plants they were collected from, and identified to species level.

For weed data, the mean coverage in each 1 m² quadrats was determined by averaging the plant values from 10 × 10 cm plot. Next the inter-annual differences in coverage were tested using multivariate analysis of variance (MANOVA) and mean coverage values obtained for one 1 m² quadrat (40 data / field type / collection dates) were considered. No significant difference in weed coverage was detected between years (P = 0.12). Therefore, data from the two years, collected on the same dates, were combined for the analyses. The weed frequency data were tested using Poisson-distributed errors and residuals for normality of errors (Kolmogorov-Smirnov test) and for equality of variance (Levene’s test). Because residuals did not meet the assumption of normality, we used the non-parametric Kruskal-Wallis- and Mann-Whitney U test to compare variables. Weed species and management systems (HIF vs LIF) were used as fixed factors and the average weed coverage in 1 m² sub-transect as random factor.

All aphid species were correlated with particular weed species. In the case of one individual weed plant hosting two aphid species, the percentage of the species were considered. This was the case in only 7% of all the samples examined. It was then determined how cropping system differentially affected associational susceptibility to the two aphid species, B. helichrysi and A. solani. General linear modelling was used with mean aphid abundance on S. annua, E. canadensis and S. canadensis as response variable. Initial analyses indicated no difference (P < 0.23) between study years and aphid abundance averaged across study years were analysed. The model included cropping system type (HIF vs LIF), aphid species (B. helichrysi and A. solani) and their interaction as explanatory variables. Because aphid abundance is a discrete variable, Poisson-distributed errors were assessed. Aphid abundances on S. annua was normally distributed, so factorial ANOVA was used, followed by Tukey testing. Aphid abundance on E. canadensis and S. canadensis did not meet the assumption of normality, hence the Kruskal-Wallis test was used, followed by the Mann-Whitney U test. Significant (P < 0.05) interactive effects (cropping system type x species) suggest that the effect of cropping system depended on aphid species. All analyses were made using R version 3.0.1 (R. Core 2013). Only a small number of other aphid species (e.g. Macrosiphum spp.) were detected, and we did not include them in the analyses.

POD activity values obtained were compared between years (MANOVA) considering values from 1 m² sub-transects per sampling period. No significant year effect was detected (P = 0.61); therefore, data between years were averaged. Thereafter, POD values were compared between cropping system type (HIF vs LIF) for S. annua. This was done because only this weed was present during the whole vegetation period in all areas in both years, while the abundance of the other weed species sampled decreased in the HIF regime; thus the POD enzyme data were not compared statistically. POD values at 5 and 10 μmol min⁻¹ · mg protein⁻¹ unit were analysed separately and compared between fields from the first to the last day of sample collection. Because the residuals meet the assumption of normality, a complete randomized factorial ANOVA of POD specific activity values (5 and 10 μmol min⁻¹ · mg protein⁻¹ unit) was performed to test for effects of treatment (HIF vs LIF) and time (data collections). The analysis of Tukey test with P < 0.01 and LSMEAN (Minimal quadratic means) according to statistic package SAS were included and the average POD quantity / 10 plants / 1 m² sub-transect were used. Linear correlation between POD activity level at 5 and 10 μmol min⁻¹ · mg protein⁻¹ unit and aphid (B. helichrysi and A. solani) abundances on S. annua plants under low and high input management were computed using the SPSS package version 3.14. Correlations were made between data (POD and aphids abundance) of the same sampling periods, and r and P values computed.

Effects of weed plants on B. helichrysi colonization toward each crop plant (maize, potato and alfalfa) were tested using repeated measures MANOVA. Interactions were then compared using χ² tests on the differences between the covariance matrices and by the root mean square error of approximation. The initial comparison was made between the two aphid species (A. solani and B. helichrysi) when these were on E. canadensis. Because of low density of A. solani, comparisons were made separately for those blocks where both aphid species were present (five blocks of E. canadensis), and for those where only the B. helichrysi persisted. The next analyses followed the comparison between weed species when these served as host plant for B. helichrysi only. Comparison between aphids densities found on crop plants (maize, potato and alfalfa) when these were set in blocks with E. canadensis or S. annua were made using Student’s t-tests, following the t-distribution. The statistical analyses were performed in R version 3.0.1 (R Core 2013).

Three invasive weed species were dominant during the two years field assessment. Stenactis annua was the most frequent weed, and dominated both LIF (97.5%) and HIF regimes (84.5%). Two other invasive weed species were present at lower densities. S. canadensis was only present in LIF, with an average coverage of 2.5%. No other invasive weeds were detected under this management system during the assessment E. canadensis was only present under HIF with an average coverage of 15%. Other weed species, mostly amaranth, Amaranthus spp. in HIF regimes with an average coverage of 0.5%, were detected during the end of the vegetation of the previous weed species. Dominance of S. annua was significant under both LIF and HIF (Table 2).

The two important aphid species were detected in high densities on all three dominant invasive weeds. The most frequent species was B. helichrysi; its abundance was high and dominated the most frequent weed, S. annua under both HIF and LIF regimes (LIF-B. helichrysi and LIF-A. solani F_1–40 = 6.4, P < 0.001; LIF-B. helichrysi and HIF-A. solani F_1–40 = 8.1, P < 0.001) (Fig. 2). The next most abundant aphid was A. solani, also present on S. annua plants under both management systems; its density was significantly higher under LIF compared with HIF (F_1–40 = 8 P < 0.001) (Fig. 2). Higher density of B. helichrysi was detected on S. canadensis under LIF (U_1–40 = 3.4, P < 0.01) but its density varied greatly between assessment data (Fig. 3A). Furthermore, the dominance of the B. helichrysi on E. canadensis was detected under HIF conditions (U_1–40 = 3.1, P < 0.01) (Fig. 3B). A very low number of other important aphid species were detected, i.e. about 12 individuals of Macrosiphum spp. were collected on S. canadensis.

Figure 2.

The average B. helichrysi and A. solani aphid densities on S. annua in LIF and HIF. Aphids from 1 m² sub-transect (cumulated and averaged between years) were considered by factorial ANOVA, followed by Tukey testing. Arrows on bars represent standard errors; different letters indicate statistically significant differences at P < 0.01 level.

Figure 3

A, B Average B. helichrysi and A. solani aphid densities on S. canadensis and E. canadensis in LIF (A) and HIF (B) field. Aphids from 1 m² sub-transect (cumulated and averaged between years) were considered by Kruskal-Wallis test followed by Mann-Whitney U test. Stars above boxplots indicates statistical significant differences at P<0.01 level.

No observable differences in POD enzyme activity were detected for S. annua at 5 μmol min⁻¹ · mg protein⁻¹ unit between HIF and LIF regimes (F_1–40 = 1.2, P < 0.2) (Fig. 4A). When the POD activity was compared for the 10 μmol min^–1 · mg protein^–1 unit aliquot sample, there was a significantly higher enzyme activity, suggesting a significantly higher stress by aphids feeding on S. annua in LIF system (F_1–40 = 3.8, P < 0.004) (Fig. 4B). Higher POD enzyme activity at both 5 and 10 μmol unit was detected on E. canadensis than on S. canadensis, again indicating higher stress as a result of aphid feeding; however, because of low samples numbers no statistics were performed here. There was a strong negative relationship between POD level at both 5 and 10 μmol unit and aphid density (both B. helichrysi and A. solani) abundances on S. annua. No such strong correlation between POD level and aphids abundance for HIF was observed (Table 3).

Figure 4

A, B POD enzyme activity at 5μmol (A) and 10 μmol min⁻¹ · mg protein⁻¹ unit (B) on S. annua and its variation between LIF and HIF. Analysis of Tukey test with P < 0.01 and LSMEAN (Minimal quadratic means) were used and the average POD quantity / 10 plants / 1 m² sub-transect were considered. Bars represent standard errors. Different letters indicate statistically significant differences at P < 0.01 level.

The number of A. solani were low and colonies persisted in five blocks on E. canadensis only, which shows a very similar trend with field observations of only 7% of A. solani detected together with B. helichrysi. The B. helichrysi colonies persisted in all blocks on both weed plants. Therefore, comparisons were made separately for those blocks where both aphid species were present, and separately for those where only the B. helichrysi persisted. Colonization of B. helichrysi from E. canadensis was significant on all crop plants, with a higher number of aphids detected on maize. Low or no colonization of A. solani was detected from this weed to crop plants, hence no statistics were here possible (Fig. 5). By comparing the colonization of B. helichrysi from both weed species, again a significant effect toward all crop plants was detected (Fig. 6). The number of aphids on maize was significantly higher (F_1–14 = 5.8, P < 0.01) when maize was in close vicinity with S. annuus. No differences in aphid abundance were detected for potato (F_1–14 = 2.5, P < 0.28) and alfalfa (F_1–14 = 1.5, P < 0.57) when these plants were in close vicinity with S. annua or E. canadensis (Fig. 6, bar charts).

Figure 5.

Colonization rate of B. helichrysi and A. solani from E. canadensis towards maize, potato and alfalfa plants. The average number of aphids / plots / plants / sampling data were considered. Interactions were then compared using χ² tests on the differences between the covariance matrices, and by the root mean square error of approximation. Numbers represent χ²-values for significant path coefficients. *P < 0.05, ***P < 0.001. Because of low A. solani numbers, no statistics were possible.

Figure 6.

Comparison between colonization rate of B. helichrysi from E. canadensis and S. annua towards each crop plant (maize, potato and alfalfa). The average number of aphids / plots / plants / sampling data were considered. Interactions were then compared using χ² tests on the differences between the covariance matrices, and by the root mean square error of approximation. Numbers represent χ²-values for significant path coefficients. **P < 0.01, ***P < 0.001. Bar chart represent comparison between aphids densities found on crop plants (maize, potato and alfalfa) when these were set in blocks with E. canadensis or S. annua. To compare variables, t-tests were used. Different letters indicate statistical significant differences at P < 0.01 level. Error bars = ±1 standard error.