Daktulosphaira vitifoliae Fitch (grape phylloxera, Phylloxeridae) is one of the most devastating pests in viticulture, mainly because of their root feeding activity. Up to today fundamental relations between belowground habitat and population dynamics remain unclear. In this 2-year study, we investigated the relations between grape phylloxera population and phenotypic traits of roots and feedings sites.
We extracted root and soil samples frequently of two closely related mature rootstocks [Vitis berlandieri × V. riparia (cvs 5C and 125AA)]. We quantified and characterised larval population and crowding and assessed root gall pigmentation, root morphology and soil parameters. We identified and described three stages of seasonal population dynamics: reproduction, overwintering and declining. Further, we demonstrated a significant impact of root gall pigmentation and crowding on population dynamics.
In temperate climates, grape phylloxera is able to overwinter at high density on roots of American rootstocks. Population dynamics are highly dependent on root gall development and the ability to crowd.
The results of our study are considered to have a significant impact on the development of management strategies for grape phylloxera.
Grape phylloxera (Daktulosphaira vitifoliae Fitch) is a root and leaf-sucking aphid (Phylloxeridae) on grapes (Vitis ssp.), native in North America (Wapshere and Helm 1987, Downie et al. 2001). Because of its worldwide spread via plant material in the late 19th and early 20th century, D. vitifoliae has turned into one of the most devastating pests in viticulture and is still causing large losses worldwide (Biada-Miro et al. 2010, Powell et al. 2013). Grape phylloxera has a variety of life cycles, ranging from cyclic parthenogenesis with temporal polymorphism and facultative sexual reproduction (Downie and Granett 1998) to a supposed obligate holocyclic asexual reproduction on leafs and/or roots of Vitis ssp. (Forneck and Huber 2009). One of the most common management strategies to control grape phylloxera is grafting tolerant rootstocks to the susceptible grape producing scions. Root feeding morphotypes of D. vitifoliae, however, are predominant in grafted vineyards (Powell et al. 2013). Abiotic factors such as temperature (Turley et al. 1997), and biotic factors such as host cultivar (Granett et al. 2007) and D. vitifoliae genotypes (Herbert et al. 2010) are impacting population performance of root-infesting grape phylloxera. Roots of Vitis ssp. are nutrition source, mating and oviposition site and the location of larval development for root feeding D. vitifoliae. Individual D. vitifoliae spend most of their live-time on the feeding sites (nodosities). Consequently, the individual root as well as the individual nodosity are the main habitat of the root-feeding morphs of grape phylloxera (Forneck and Huber 2009). Questions remain as to whether and how the dynamics of habitat conditions impact grape phylloxera performance and eventually population development. Although some studies focused on population performance of D. vitifoliae on the roots of Vitis in field (Helm et al. 1991, Omer et al. 1997, Granett et al. 2001, Porten and Huber 2003), to our knowledge, no recent study has continuously monitored the population dynamics and root feeding site conditions, using modern rootstock cultivars. Particularly, factors such as crowding (Agrawal 2004), root development and feeding site development (Bauerle et al. 2007) may play an important role in population dynamics. In this study, we continuously investigated grape phylloxera populations on roots of two Vitis riparia × Vitis berlandieri rootstocks (cvs 5C and 125AA). We assessed the larval development of root parasitising morphs of grape phylloxera and the phenotypic development of their feeding sites over 2 years under field conditions.
We used a commercial vineyard, located in Geisenheim, Germany (GPS: N 49.996054, E 7.951049; 157 m absolute altitude). Two V. berlandieri × V. riparia rootstock cultivars (cvs 5C and 125AA), grafted with Vitis vinifera Weisser Riesling, were investigated for 2 years. The study site was planted in 1985 and is conventionally managed, with application of organic fertiliser every other year. During the field investigations of this study, the vineyard was fertilised with cow manure in April 2008 (40 kg/ha).
Root and soil samples were collected periodically in the 20 months between August 2007 and August 2009 (Table S1). Samples were taken randomly in equal amounts per variant (5C, 125AA). All samples were collected directly under a plant, and each plant was sampled once during the whole study.
The roots and attached D. vitifoliae were extracted with a quadrangular metal box (0.1 × 0.1 × 0.2 m) from the top 20 cm of soil. Roots were roughly cleaned in the field and stored in water at 7°C for transportation. Prior to the sampling (June–December 2006), the sampling method employed was compared with the field assessment methods published in Porten and Huber (2003) with no significant differences through summer months and higher numbers of individuals in winter (Table S2, Figure S1). To assess abiotic and biotic soil conditions, separate soil samples were taken using an Oakfield tube soil sampler in the top 20 cm and stored in plastic bags at 7°C for transportation. Soil samples were stored in the laboratory at 4°C and processed within the following 2 weeks after field work. Root samples were investigated under a microscope for the population of D. vitifoliae and gall formation within the following days after field sampling. After the visual survey for grape phylloxera, root samples were gently washed under tap water, and the cleaned roots were used to assess root morphological parameters via Win Rhizo Pro V2005b (Regent Instruments, Nepean, ON, Canada).
Root-infesting individuals of D. vitifoliae were assessed according to Hoffmann et al. (2011). Grape phylloxera instars were classified into juvenile wingless [larval stage 1–4 (L1–L4); non-oviparous], adult wingless [larval stage 5 (L5); oviparous] and nymphal larvae (prewinged; non-oviparous) (Table 1).
We used three parameters to indicate population density on grape roots: (i) population density was calculated per cm2 root surface area; (ii) the mean number of insects per root gall was calculated for each assessed class of larval stage and for each assessed root gall class; and (iii) the mean number of conspecific neighbours per occupied root gall (crowding) was calculated for every assessed class of larval stages (Table 1).
After assessing the grape phylloxera population, root galls were characterised into three classes, following their pigmentation (Hoffmann et al. 2011). The classes were as follows: ld, low dark pigmentation; md, medium dark pigmentation; and hd, high dark pigmentation. The colour classes were assessed before by digital analyses of extracted roots with Win Rhizo Pro V2005b. Additionally, the position of a root gall in the root system (terminal, not terminal) was recorded.
Two morphological root system parameters were assessed after the visual survey of roots for insects and root galls: average root diameter (mm) and root surface area (cm2). The cleaned root samples were digitised by creating greyscale images of all root fragments, using the transmitted light unit of a Perfection 4990 Scanner (Epson, Suwa, Japan). The morphological structure of the roots was determined by analysing the images with Win Rhizo Pro V 2005 beta. To compare assessed morphological parameters to root mass, the dry mass of the roots was assessed.
Four soil parameters were investigated: soil water content (SW), pH, soil organic matter and basic respiration (BR) (Table 1). All soil samples were sieved through a 5-mm mesh. To determine SW, the proportion of the dry sample mass (105°C/24 h) was calculated [dry mass of soil (%)]. Soil pH value was determined by suspending 10 g of soil sample in 25 mL of 0.01 mol/L CaCl2 solution. Soil pH value was measured with a pH metre after 30 min. Soil organic matter was assessed by detection of the loss of ignition after the incineration of the soil samples. To produce ash, dried (105°C/24 h) soil samples were incubated in a muffle furnace (400°C/24 h). Loss of ignition was assessed by determining the ash mass proportion (%). The biotic parameter BR was measured according to Isermeyer (1952). Basic respiration is a parameter related to the activity of CO2 producing soil microorganisms and was determined by adjusting 2 g soil sample to 40% moisture content with sterile demineralised H2O. Each sample was placed in a glass box (1000 mL), containing a beaker with 2 mL ∙ 1 mol/L NaOH. The containers were closed airtight at 22°C for 24 h in the dark. In NaOH solution, the CO2produced binds as Na2CO3. To eliminate Na2CO3, BaCO3 (saturated) was added to the solution after 24 h (fall out of BaCl2). The mass of untransformed NaOH was determined by a volumetric analysis (titration) with HCl (1 mol/L) (tracer: phenolphthalein), and BR is specified in milligrams carbon per gram soil dry mass and hour [mg C/(g ∙ h)].
Environmental parameters for the location Geisenheim were provided by the German Bureau for Weather Forecasts, Deutscher Wetter Dienst, Offenbach, Germany. Coordinates of the measuring unit are 49°58′59″N, 7°57′00″E. In this study, we calculated monthly mean values of air temperature based on daily mean values of temperature. Furthermore, we calculated the monthly sum of precipitation based on daily sums of precipitation (mm). Monthly overview of temperature and precipitation for the period of this study (2007–2009) is provided in Figure S2.
Density of grape phylloxera instars, root galls and root system parameters was described as mean values (± standard error). We assessed general differences in seasonal development of grape phylloxera population and root gall development by performing multiple ANOVAs for the density of each class of larval stages and root galls (fixed effect models, factors: season and variant, P = 0.05). One-way ANOVAs were performed (fixed effect model, factor: season, P = 0.05) to investigate seasonal rootstock (125AA, 5C) dependent differences in grape phylloxera population development. In case of significance, Tukey honest significant difference (HSD) post hoc tests (P = 0.05) were calculated for each group (rootstock variant and season).
To describe the effect of habitat and crowding on grape phylloxera population development, the assessed parameters were grouped into four categories: (i) conspecific neighbours; (ii) root galls; (iii) root system parameters; and (iv) soil and weather conditions. A principal component analysis was performed on each of those categories [FactoMineR (Lê et al. 2008)]. First and second principal components (PC1 and PC2) were extracted. The values of the principal components were used as variables to explain population development in a generalised linear model (all instars per cm2 − PC1 × PC2; using a log link function for Poisson distribution). Correlations were performed between PC1/PC2 and the parameters included in the categories.
Graphical illustrations and descriptive statistics were calculated in Excel 2010 (Microsoft, Redmond, WA, USA). All analytical calculations were done in R 3.0.2 (www.r-project.com).
The highest density of grape phylloxera instars occurred in mid-summer (21 July 2008 and 13 July 2009, average of 1.69 individuals per cm2 root surface). The lowest grape phylloxera population density was measured in late spring (15 May 2008 and 9 May 2009, average of 0.008 individuals per cm2), together with the lowest density of root galls (0.31 per cm2) (Figure S3). Adult wingless and nymphs mainly occurred during summer months (Figure S4). Grape phylloxera population density per cm2 root surface differed significantly between rootstock variants. Oviparous adult wingless (L5 per cm2 root surface) occurred significantly more often (P = 0.03, Tukey HSD) in summer on 5C than on 125AA (Figure 1a). Significantly (P < 0.01, Tukey HSD) fewer juvenile wingless (L1–L4 per cm2 root surface) occurred in spring compared to all other seasons (Figure 1b). No differences could be detected between rootstock variants considering the occurrence of nymphal larval stages (Figure 1c). The population of juvenile wingless was unexpectedly stable overwinter on 5C. Although a significant decrease in the occurrence of juvenile wingless (P = 0.02 Tukey HSD) could be detected on 125AA overwinter, it was not observed on the roots of 5C (Figure 1d).
The mean number of juvenile wingless per root gall was highest in January (0.69 individuals on ld galls) and lowest in March (0.05 individuals on ld galls). The number increased again in late summer to a second peak in October (0.61 individuals per ld root gall) (Figure S5). The number of individuals per root gall was significantly lower in spring (P < 0.01, Tukey HSD) than in all other seasons of the year, including winter (Figure 2a). The occupation of root galls was reflecting population density per cm2 root surface (Figure 1b).
Significant crowding effects could be observed in autumn and winter months. The number of conspecific neighbours increased significantly compared to spring and summer (P < 0.01, TukeyHSD; Figure 2b). Significant differences between 125AA and 5C in crowding activity were observed in autumn (P = 0.01, Tukey HSD). The mean number of juvenile wingless as conspecific neighbours of juvenile wingless was lower than 0.5 in spring and summer and higher than 1 in winter and autumn.
No significant difference in root surface development and root gall development could be measured between rootstock variants (data not shown). The mean root surface increased from a winter average of 47.7 cm2 per 2 L soil volume to an average of 97 cm2 per 2 L soil volume in summer (Figure S3). The main surface area development occurred in roots between 0 and 1 mm diameter (Figure 3a). The mean root surface of small diameter roots (0–0.5 mm) doubled significantly from winter (~15 cm2 in 2 L soil) to summer (~30 cm2 in 2 L soil) and was decreasing from summer to winter again. In average, root galls occurred three to four times more often than insects (per cm2 root surface). Root surface and the density of root galls with low pigmentation had their lowest level in winter.
Over all rootstock variants, density of ld pigmented root galls was significant lower (P < 0.001, Tukey HSD) in winter and spring than in summer and autumn (Figure 3b). The highest level of root galls with ld pigmentation was recorded in August, parallel to the peaks in the occurrence of oviparous adult wingless and nymphal larval stages (Figure S3). The highest total occurrence of root galls, however, was measured in winter with 0.52 root galls per cm2root surface, because of significant higher density of md (P = 0.04, Tukey HSD, Figure 3c) and hd pigmented root galls (P = 0.02, Tukey HSD, Figure 3d). The number of root galls with continuous root growth (non-terminal) was not significantly different to the number of terminal root galls (data not shown).
The soil water content decreased from winter (18.6% ± 0.2) to summer (11.09% ± 0.2), whereas soil organic matter, soil pH and BR remained stable between seasons (Table S3). Mean temperature had a minimum in January (3.2°C in 2008 and -0.9°C in 2009) and a maximum in July 2008 (22.3°C) and in August 2009 (23.7°C). Lowest precipitation rate occurred in October 2007 (6.3 mm) and in August 2009 (11.2 mm), high rainfall was observed in June 2008 (78.7 mm) and in July 2009 (76 mm) (Figure S2).
Principal component analyses were performed in four different categories: (i) conspecific neighbours; (ii) root galls; (iii) root system; and (iv) soil and abiotic parameters (Table S4). Afterwards we used PC1 and PC2 as variables to explain grape phylloxera population per cm2 in a general linear model (GLM) (log link for Poisson distribution). In groups 3 and 4, two dimensions explained 60.86 and 56.84%, respectively, of total variance. In the performed GLMs, however, no significant impact of root morphology, soil and abiotic factors could be observed (Table 2). In group 1 (conspecific neighbours), two dimensions explained 46.98% of total variance (PC1, 31.4%; PC2, 15.58%) with PC1 having a significant impact on the explanation of total grape phylloxera density in the GLM (Table 2). All factors were positive correlated to PC1 (Figure S6), whereas the conspecific neighbours of nymphal larval stages were negatively correlated to PC2 (Figure 4b). In group 2 (root galls), two dimensions explained of 66.82% of variance (PC1, 44.77%; PC2, 22.05%; Figure 4a). Both components (PC1 and PC2) had a significant impact on the explanation of total grape phylloxera density in the GLM (Table 2).
We observed three stages of seasonal development of root-infesting grape phylloxera population (Table 3). The reproductive stage lasted from early summer to mid-autumn. It was characterised by a diverse population structure (juvenile wingless, adult wingless and nymphal larval stages), high population density, lower crowding rate and a light pigmentation of feeding sites. The overwintering stage lasted from late autumn to early spring. It was characterised by stable but slightly lower population density with only wingless juveniles occurring. The crowding activity in the overwintering stage was significantly higher than that in the reproduction stage, and significantly fewer light pigmented feeding sites were observed. We also observed the third-stage population decline, which lasted from early spring to early summer and was characterised by decreasing population density and no oviparous larval stages, low crowding activity and a low level of light pigmented feeding sites. These observations are partly in accordance with observations made earlier. Omer et al. (1997) and Porten and Huber (2003) observed similar patterns in grape phylloxera population dynamics. Although knowledge of grape phylloxera population dynamics in the field is crucial for grape phylloxera management (Benheim et al. 2012), the work of Omer et al. (1997) is the only study so far in which grape phylloxera populations were recorded overwinter in California vineyards, occurring in low population density [per root dry mass (dw)]. In contrast, we observed a high density of overwintering juvenile wingless (per cm2root surface), indicating a different population development overwinter in temperate climate viticulture. Significant differences between overwintering habits of grape phylloxera were already described, for example, by Troitzky (1929) who investigated field populations between Moldavia and Azerbaijan. The difference between our observations of high overwintering rates and the low insect density observed by Omer et al. (1997) could be explained by several factors. Omer et al. (1997) did not investigate population density on V. berliandieri × V. riparia rootstocks. On V. vinifera × Vitis rupestris AXR#1, in a vineyard in a cooler climate, they observed a maximum of around 40 L1–L4 per g root dry mass. This was about half the observed population density on own rooted V. vinifera, observed also by Omer et al. (1997) in a warm, irrigated vineyard. One gram of root dry mass is equal approximately between 70 and 130 cm2 of root surface (Figure S7). A population density of 0.02 individuals per cm2 root surface (observed in winter) would be equal to 1.4–2.6 individuals per g root dry mass, a density of 0.14 individuals per cm2 root surface (observed in summer) would be equal to 9.8–18.2 individuals per g dry mass. These numbers are lower than the population density found by Omer et al. (1997). Besides seasonal, rootstock-related and environmental factors that impact grape phylloxera development, the difference in genotype or biotype [see e.g. Herbert et al. (2010)] between the investigated populations by Omer et al. (1997) and the populations investigated in this study could be an additional reason for a difference in population density. Although we observed seemingly a low population density in summer, Huber et al. (2009a) showed that even marginal root damage caused by grape phylloxera can lead to infections of the plant with secondary fungal pathogens.
The production of cold-adapted hibernating morphs, however, is known from other root-infesting aphids such as the lettuce root aphid Pemphigus bursarius (Judge 1967, Moran 1992). Early authors, such as David (1875), who observed living grape phylloxera on grape roots in 20 cm depth of frozen soil, or Cornu (1878), who described a continuous appearance of nodosities over several years in temperate vineyards in France, were suggesting high overwintering rates in temperate viticulture on roots of susceptible V. vinifera. In this study, we demonstrated a significant increase in crowding activity of overwintering grape phylloxera populations also on tolerant American rootstocks. One explanation for the observed crowding activity could be a change in movement behaviour of grape phylloxera. Turley et al. (1997) observed increased movement behaviour in root feeding, first instar juvenile wingless grape phylloxera, but is suggesting reduced movement of grape phylloxera in cool temperature, assessed by excised root assays. Both increased and decreased movement behaviour, however, could lead to higher crowding rates on feeding sites. Factors and mechanisms that are involved with grape phylloxera movement and attraction are not yet known for certain yet [see Powell et al. (2013)]. The relevance of intrinsic triggers, however, on aphid population dynamics were reviewed by Kawada (1987) for example. A variety of both extrinsic and environmental factors are discussed in grape phylloxera-specific literature (Omer et al. 1997, Porten and Huber 2003, Forneck and Huber 2009). Omer et al. (1997), for example, suggest a reduced direct influence of environmental conditions such as temperature on grape phylloxera population dynamics. Because we did not record a relation between temperature and grape phylloxera density in field, we agree with Omer et al. (1997) and suggest extrinsic factors other than temperature are important for grape phylloxera field dynamics. Considering phylloxera evolution and its distinct host specificity, it is most likely that there is a strong host–plant-mediated effect on phylloxera life cycle (Dixon 1987). As grapes are the only host plant of grape phylloxera, changes in nutritional value and chemical composition of the food source may be a crucial factor in the determination of phylloxera life cycle stages. Nutrition availability and composition play a role in attraction of herbivorous insects (Dill et al. 1990). Because of a lower concentration of starch in roots in winter (Stoev et al. 1966), starch accumulation might occur less often and consequently would be more attractive for grape phylloxera in winter. Starch delocalisation from root to shoot could also be an explanation for the observed declining stage in spring. Other compounds are also known for their variability in plants during the vegetation period or between plant tissues such as amino acids (Rilling et al. 1974) or brassinosteroids (Bajguz 2011). Additional modification of this plant-mediated life cycle determination may be due to microbial root and soil community structure (Huber et al. 2009a) or endosymbiotic associations (Vorwerk et al. 2007). Observations to such effect were already made, for example, by Balbiani (1874) and Cornu (1878) who observed that the development of grape phylloxera instars is related to local root and soil conditions. But phylloxera intrinsic and host plant-mediated extrinsic factors are only part of phylloxera life cycle determination and behaviour, especially considering overwintering and crowding. Phylloxera intrinsic and host plant-mediated extrinsic effects may essentially determine induction of phylloxera life cycle stages and morph determination; there is ample evidence that host finding and detection of suitable feeding and overwintering sites are influenced by the host plant and phylloxera sensory perception. Lawo et al. (2011) identified a significant increase of plant defence-related volatiles in nodosities of 5C rootstocks. But it is not clear whether these volatiles impact grape phylloxera dispersal. From other systems, it is known that defence volatiles can attract or repel plant feeding insects [Mithofer and Boland (2012)]. A direct effect of phylloxera population structure on dispersal and aggregation was shown by Clever (1959a,1959b) whose findings indicate that chemotaxis may be involved in finding of feeding sites and between phylloxera individuals. The general ability for sensory perception of different external stimuli of grape phylloxera was shown by Huber et al. (2009b) who identified chemoreceptors in the antennal sensilla of grape phylloxera.
Seasonal grape phylloxera population density varied significantly between rootstocks. Generally 5C supported a higher amount of grape phylloxera than 125AA. Generally, grape phylloxera population performance differs with rootstock cultivars [see Powell et al. (2013)], and in the laboratory, Kocsis et al. (1999) observed high performance of grape phylloxera on 5C. We recorded a significantly lower density of oviparous larval stages on 125AA, resulting in a significant lower density of juvenile wingless over winter. In contrast, we did not record any significant difference in root gall density between rootstocks, indicating different efficiencies of plant defence mechanisms between 5C and 125AA after nodosity formation! A low level of defence efficiency in 5C could also be an explanation for the observed significant higher crowding activity on roots of 5C. Generally, the aggregation of herbivorous insects depends on nutrition content as well as on host species (Agrawal 2004). Crowding can have an impact on insect physiology. Especially larval development time can be increased (Agrawal 2004) or decreased (Sillanpää 2008) by aggregation. In our study we demonstrated that a higher population density is related to a higher number of conspecific neighbours per root gall. Especially the overwintering of juvenile wingless depends at least partly on the ability to crowd on nodosities.
Leaf-infesting morphs of grape phylloxera are inducing stomata and local metabolic changes in leafs (Nabity et al. 2013). To date, fundamental mechanisms related to nodosity formation are mostly unknown. Our results indicate, however, that grape phylloxera population development depends on the ability to induce and to stay on root galls. Moreover, the ability to crowd on root galls seems to be crucial for a successful overwintering of juvenile grape phylloxera.
For the first time, we quantified seasonal differences in the performance of grape phylloxera between mature rootstocks of the same group (Berlandieri-Riparia group) in a field study. The seasonal dynamics of root-infesting grape phylloxera population can be characterised by three stages: reproduction, overwintering and declining. The performance of grape phylloxera population during reproduction and overwintering depends on the rootstock cultivar. Overwintering of juvenile wingless grape phylloxera is associated with high crowding rates and could be observed predominantly on 5C. It can be assumed that the ability to crowd as well as the ability to induce nodosities is impacting grape phylloxera population dynamics. Beside plant intrinsic defence and growth factors, crowding as well as grape phylloxera survival might be partly related to extrinsic factors.
This work was funded by the German Research Foundation (DFG) and the German Foundation of Grapevine Research (FDW). We particularly thank Mr Franz Boehm for the permission to take root and soil samples in his vineyards. We further thank Dr Xavier Martini and Dr Monique Coy (both University of Florida) for critical remarks and discussion of the manuscript.
