J. Ent. Acar. Res. Ser. II, 42 (2): 103-116 30 August 2010 Z. P. GAMA, P. MORLACCHI, A. GIORGI, G. C. LOZZIA,  J. BAUMGÄRTNER Towards a better understanding of the dynamics of Aphis spiraecola Patch (Homoptera: Aphididae) populations in commercial alpine yarrow fields  Abstract - The spatial distribution of Aphis spiraecola Patch was studied in two  commercial yarrow fields located in the Swiss and Italian Alps and represented by  Taylor’s (1961) power law. The respective parameters indicate a highly aggregated  distribution and lead to a high optimum sample size of 400-500 plants in the design  of a sampling program. Opportunities for reducing the sampling efforts are discussed.  The infestation patterns were studied on the basis of Vansickle’s (1977) time varying  distributed delay adequate for modelling the dynamics of age-structured populations.  Published literature data were used to parametrize the functions representing the  temperature-dependent duration and survival of the nymphal and adult stage.  Likewise, literature data were available to obtain reliable estimates for the parameters  of the fecundity function comprising the reproductive profile and the number of  nymphs produced at different temperatures. The field data were used to parametrize  the functions for wing formation and a compound mortality compromising the effects  of plant senescence, stem cutting and natural enemies. The model satisfactorily  represented the observed infestation patterns. However, there are opportunities for  improving parameter estimation and validation. Moreover, the separation of the  compound mortality into host plant and natural enemy effects would improve the  mechanistic basis of the model and lead towards a tool that could be used to study  bottom-up and top-down effects in the yarrow-aphid-natural enemy system. Riassunto - Verso una migliore comprensione della dinamica di popolazioni di Aphis spiraecola Patch (Homoptera: Aphididae) in campi commerciali alpini di  Achillea. È stata studiata la distribuzione spaziale di Aphis spiraecola Patch in due campi  coltivati di Achillea situati sulle Alpi italiane e svizzere; la distribuzione spaziale  della specie è stata descritta dalla legge della potenza di Taylor (1961). I parametri  specifici indicano una distribuzione spaziale fortemente aggregata e nel contesto  di un programma di campionamento portano al calcolo di un’elevata dimensione  ottimale del campione, pari a 400-500 piante. Le possibilità per una riduzione  dell’impegno per il campionamento vengono discusse. Journal of Entomological and Acarological Research, Ser. II, 42 (2), 2010104 Gli schemi di infestazione sono stati studiati sulla base del modello a ritardo distribuito  a tempo variabile di Vansickle (1977), adatto per le dinamiche di popolazioni  strutturate per età. I dati di letteratura sono stati usati per la parametrizzazione delle  funzioni che descrivono la durata e la sopravvivenza temperatura-specifiche degli  stadi di sviluppo preimmaginali e immaginali. Allo stesso modo, i dati di letteratura  sono stati impiegati per ottenere stime affidabili per i parametri della funzione per la  riproduzione, che comprende il profilo riproduttivo e il numero di neanidi prodotte  a differenti temperature. I dati di campo sono stati utilizzati nella parametrizzazione  delle funzioni formazione degli individui alati e mortalità complessiva, composta  dai fattori senescenza della pianta, taglio del culmo e nemici naturali. Il modello  descrive in modo soddisfacente gli schemi di infestazione osservati. Tuttavia esistono  possibilità per il miglioramento della stima dei parametri e della validazione. Inoltre,  la separazione della mortalità complessiva nelle componenti pianta ospite e nemici  naturali migliorerebbe la base meccanicistica del modello e indirizzerebbe verso uno  strumento che potrebbe essere usato nell’analisi degli effetti bottom-up e top-down  nel sistema achillea-afidi-nemici naturali. Key words: Aphis spiraecola, Achillea collina, spatial distribution, sampling plan,  infestation pattern, delay model, parameter estimation, model validation  INTRODUCTION Yarrow (Achillea collina Becker ex Rchb.) is cultivated for commercial purposes  in the European Alps. Of interest in human medicine is the high content of secondary  metabolites, i.e. organic compounds that are not directly involved in the normal growth,  development, or reproduction of organisms (Fraenkel, 1959; Wink, 2003; Madeo et al.,  2009). The aqueous and alcoholic extracts have digestive, antiphlogistic, spasmolytic,  stomachic, carminative, and estrogenic properties (Benedek et al., 2007). In two fields  located in the Southern Italian and Swiss Alps, Morlacchi et al. (2010) studied crop  yield formation and recorded two insect communities of possible economic importance.  The first community, not studied in this paper, consists of three leaf eating chrysomelids  (Galeruca tanaceti L., Chrysolina marginata marginata L., Cassida spp.) and their  natural enemies. The second community comprises three phloem feeding aphid species  (Macrosiphoniella millefolli DeGeer, Aphis spiraecola Patch and Coloradoa achilleae Hille Ris Lambers), their parasites and the predator Coccinella septempunctata L. The  populations of these aphids occurred in sufficiently high numbers as to possibly reduce the  yield in terms of biomass on one hand and increase the contents of secondary metabolites  on the other hand (Madeo et al., 2009). An adequate knowledge on the spatio-temporal  dynamics of the aphid populations is indispensable for the design of an integrated pest  and crop management system (Gutierrez and Baumgärtner, 2007).  The Spirea aphid A. spiraecola of interest in this paper is a polyphagous species  of Far Eastern origin with a worldwide distribution (Wang and Tsai, 2000). It is a pest  of citrus, apples and ornamentals, and transmits a number of plant viruses (Wang and  105Z. Penata Gama et al.: Dynamics of A. spiraecola in commercial alpine yarrow fields  Tsai, 2000). For supervised control purposes, Hermoso de Mendoza et al. (2006) defined  intervention thresholds in Spanish citrus orchards. For rationalizing control, Hong et al. (2003) identified the sex pheromone and studied the circadian rhythm in release.  The density-dependent effect of predators on A. spiraecola in apple orchards (Brown,  2004) stimulated attempts to manage A. spiraecola populations by enhancing biological  control (Brown and Matthews, 2008). In a classical biological control effort, the aphelinid  Aphelinus gossypii Timberlake was introduced into Florida to control A. spiraecola on  citrus (Hoy and Ru, 2008).  The purpose of this paper is to describe the spatial distributions, to design sampling  plans, and to analyze, via the development of mechanistic population models, the  temporal infestation patterns of. A. spiraecola in commercial alpine yarrow fields. The  model parameters are estimated on the basis of published life table data (Wang and  Tsai, 2000), the assumed formation of winged morphs (Holst and Ruggle, 1997) and the  observed natural enemy presence (Morlacchi et al., 2010). The design of sampling plans  and the analysis of infestation patterns should provide indications for obtaining reliable  density estimates for model validation and population management purposes, and for  improving the population model with respect to its mechanistic basis.  MATERIAL AND METHODS Study sites and population sampling For the study, we selected two commercial yarrow fields located in the Southern  Alps (Poschiavo, Canton of the Grisons, Switzerland, 1140 m asl, and Dazio, Sondrio  province, Italy, 900 m asl). At both locations, the farmers planted the variety ‘Spak’,  selected by Valplantons BIO (Saillon, Switzerland) for a high content of secondary  metabolites (Morlacchi et al., 2010). At the time of the study (2007, 2008), the plants  were several years old and grown on black plastic mulch at a 0.5 m x 0.5 m spacing. At  Poschiavo, the plants were harvested on July 25 (2007) and July 30 (2008), while the  Dazio grower renounced on cutting the plants during the year of observation (2008).  This study deals with the dynamics of A. spiraecola populations inhabiting the two fields  between the beginnings of April to the ends of July.  The fields at Poschiavo and Dazio were divided into 9 and 7 strata, respectively. In  each of the strata, the beating tray method was applied to 3 randomly selected plants to  obtain the number of A. spiraecola mummies of parasitized A. spiraecola, and coccinellid  larvae and adults (Morlacchi et al., 2010). In the relatively small commercial yarrow  fields, destructive sampling was not possible.  In 2007 and 2008, the Poschiavo field was visited 8 times (April 24, May 9, May 25,  June 8, June 22, July 8, July 26, September 5) and twice (June 3, June 24), respectively.  In 2008, the Dazio field was visited three times (June 3, June 24 and July 21). Since  Morlacchi et al. (2010) did not find a significant difference between strata, the samples  are treated as simple random samples taken from a homogenous sampling universe.  Journal of Entomological and Acarological Research, Ser. II, 42 (2), 2010106 Spatial distributions and optimum sample size The sampling program provided the means, variances and standard errors of 13  samples. The ratio of the standard error to the mean is used to assess the reliability of  the estimates. The spatial distribution of A. spiraecola was described by Taylor’s (1961) power  law that expresses the variance (s2) in relation to the mean (m) by .                [1] To obtain estimates for a and b through least square linear regression techniques, equation [1] was changed into ln(s2)=ln(a) +b ln(m). The optimum sample size is the smallest number n of sample units that satisfies the  objectives of the sampling program and achieves the desired precision of the estimate. For  calculating n, we defined the reliability in terms of formal probabilistic statements with  the length D of the confidence interval equal to a proportion of the mean (Karandinos,  1976). The consideration of equation [1] yields the optimum sample size n 2 2 2/ −⎟ ⎠ ⎞ ⎜ ⎝ ⎛ = bma D z n α              [2] where zα/2 is the upper α/2 point of the standard normal distribution. The definition of the  optimum sample size n depends on the objective of the sampling program (Karandinos,  1974). The values of zα/2= 1.65 and D = 0.3 reflect a high end of a range that is considered  reasonable for pest management purposes (Hutchison et al., 1988).  Basic model If the variability in developmental time is high relative the mean developmental time,  a stochastic model may be appropriate (Di Cola et al., 1999). In this work, we use the  time varying distributed delay of Vansickle (1977) to model the development of both the  nymphal and adult cohorts. However, we limit the description of the model to the basic  elements only and refer the reader to the recent examples of Gutierrez (1996), Holst and  Ruggle (1997), Alilla et al., 2005, Severini et al. (2009), Gutierrez and Baumgärtner  (2007) and Limonta et al. (2009a;b) for additional explanations and applications. Briefly,  ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ ++−= − dt tDELd k tDEL tARtrtr tDEL k dt tdr ii i )()()(1)()( )( )( 1           [3a]   i=1,2….k  where t = time [days], ri(t) = the transition rate of the i-th sub-stage, k = number of delay  sub-stages, DEL(t) = time dependent developmental time [days] in absence of losses, and  AR(t) = time dependent proportional losses or attrition. The output rk(t) of the nymphal  107Z. Penata Gama et al.: Dynamics of A. spiraecola in commercial alpine yarrow fields  stage becomes the input x(t) into the adult stage. For constant temperatures and a cohort  input x(t) into the first sub-stage, Vansickle (1997) describes the procedures for obtaining  estimates for the parameters k, DEL(k) and AR(t) as follows. 2 2 s k μ =     [3b] ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ − = kDEL 1 εμ [3c] ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ −= DEL kAR 11 μ     [3d] where μ is the observed developmental time with s2 = variance, and ε = stage-specific  survival. The input into the larval stage corresponds to the below described fecundity  rate, while input into the adult stage corresponds to the output of the larval stage modified  by the proportion of emigrating winged aphids. The survival of larvae depends on  temperature and the compound effect of predation, parasitism, plant senescence and  cutting. These model components are described in the next section. For simulation  purposes, we select a time increment of 1h for which the mean temperature is calculated  on the basis of a cosine function fitted through the daily temperature maxima and minima  (Bianchi et al., 1990). Model components The developmental rate of nymphs and adults is represented by the model of Brière  et al. (1999) ( )βαμ TTTTTT ul −−= )()( [4] where the Tl and Tu = the lower and upper thresholds for development, α and β =  parameters. While Tu has been estimated from experiments and β has been set to 0.5  (see Brière et al. 1999), Tl and α have been estimated via linear least square regression  techniques applied to the data of Wang and Tsai (2000). The intrinsic survival e of larvae  only is represented on the basis of a Beta function ( ) ( ) ( ) ςξ λε TTTTT ul −−=     [5] whose parameters λ , ξ , and ς were estimated by applying linear least square regression  techniques to the data reported by Wang and Tsai (2000). The values are reported in Tab.  1. Following Curry and Feldman (1987), the reproduction is based on the reproductive  profile fi, i.e. the normalized age-specific fecundity rate in the i-th sub-stage, and the  temperature-dependent total fecundity F(T)  Journal of Entomological and Acarological Research, Ser. II, 42 (2), 2010108   [6]   [7] The parameters τ, υ ø φ and ι are estimated by applying linear least square regression  techniques to the data reported by Wang and Tsai (2000), while R corresponds to the  total per capita fecundity realized over the k sub-stages. The product of eqs. 6 and 7  multiplied by the i-th transition rate of equation 3a and the time step length (1/24) yields  the fecundity of the i-th age group per time step with temperature T. The reproduction  of the adults, becoming the input x(t) into sub-stage 1 of equation 3a, in all age groups,  is obtained by summing the fecundities realized in all sub-stages (i=1,2,..k). Curry and  Feldman (1987) provide further details on modelling reproduction under time-varying  temperature conditions. The proportion w(N) of winged aphid depending on aphid density (N) is derived  from Holst and Ruggle (1997) ( ) ων +−+ = Ne Nw ln1 1 )(   [8] According to the field observations reported in Fig. 1, the proportion w(N) may be  low (0.05) and high (0.99) at densities N = 20 and N = 40, respectively. This tentative  interpretation allows the calculation of the parameters ν and ω reported in Tab. 2. As  indicated above, the input x(t) into the adult stage is modified by (1-w(N)). The physiological time-dependent compound effect v(τ) of natural enemies, plant  senescence and cutting is represented by  ( ) ψτρτ +−+ = ln1 1 )( e v   [9] where τ = physiological time in day-degrees (dd) above the lower developmental  threshold Tu.. The field observations reported in Fig. 1 suggest a small value for v(τ) =  0.05 at τ = 1200 dd and high value v(τ) = 0.99 at τ =1600 dd. This tentative interpretation  allows the calculation of the parameters ρ and ψ reported in Tab.1. The product of the  stage specific survival of nymphs (equation 5) and (1-v(τ)) for both nymphs and adults  is the basis for the calculation of attrition AR(t) according to equation 3d. Model validation The validation procedures consist of testing the model capabilities with respect to  its intended use (Rykiel, 1996) which is the mechanistic representation of infestation  patterns. The predictions of the model are compared to the observations made in 2007  109Z. Penata Gama et al.: Dynamics of A. spiraecola in commercial alpine yarrow fields  and 2008 in the Poschiavo yarrow field. This field may receive higher radiation levels and  is located at a lower altitude than Robbi-Poschiavo where the temperature was recorded.  Moreover, the field is situated near a lake with presumably mitigating effects on low  temperatures. To take into account these observations, the daily temperature maxima and  minima recorded at Robbi-Poschiavo were increased by 1.0 °C, and the resulting values  were used to calculate the hourly mean temperatures. Taking into account the observed  infestations, the simulation tentatively starts with a density of 5 medium age nymphs in  sub-stage i = 35 on day 100 corresponding to 391 daydegrees after January 1st.  The infestation patterns was simulated first with intrinsic parameters only, i.e. in  absence of the effect of wing formation and compound mortalities. Second, the infestation  pattern was simulated with intrinsic parameters and wing formation but absence of  compound mortality effects. Third, the infestation patterns were simulated with intrinsic  parameters, wing formation and compound mortality.  RESULTS Fig. 1 shows the logarithm of the variance plotted against the logarithm of the  mean density for each sample. The parameters a = exp(2.6552) and b = 1.926 reported  in Tab. 1 indicate, for the sampling method used here, a highly aggregated distribution  of A. spiraecola. Fig. 1. The relationship between ln (variance) and ln (mean) density of Aphis spiraecola sampled  in two alpine yarrow fields. Journal of Entomological and Acarological Research, Ser. II, 42 (2), 2010110 Fig. 2 shows the optimum sample size, i.e the number of pants to be sampled for  estimating the densities with zα/2 = 1.65 and D = 0.3. Accordingly, to obtain reliable  density estimates for research purposes at low densities, the high number of about 500  plants needs to be sampled. This requires high investments into sampling studies and  relatively big fields to facilitate random sampling. A high proportion of samples may  require finite population corrections in statistical analyses (Cochran, 1977). To obtain  reliable estimates for population management purposes, the optimum sample size can  be reduced to the still high number of 400 plants. Fig. 2. The optimum number of plants required for estimating the density of Aphis spiraecola with a predefined level of reliability (the standard normal variate zα/2 = 1.65, and the ratio of the  standard error to the mean D = 0.3). Tab. 1 lists the parameter estimates for Taylor’s (1961) spatial distribution model  (a,b, equation 1), for the order of the Vansickle’s (1977) delay (k, equation 3), for the  developmental rate according to Brière et al. (1999) (α, β, Tl , Tu, equation 4), and for  the stage-specific intrinsic survival (λ, ξ, ς, equation 5). While the spatial distribution  is described for the combined densities of adults and nymphs, the two life stages are  separated in the study on aphid infestations. Consequently, the order of the delay and  the developmental rate parameters differ between nymphs and adults. Only nymphs  suffer from intrinsic mortalities, while adult survivorship is controlled by the order of  the delay. The order k=13 satisfactorily yields the observed adult survivorship in the  data of Wang and Tsai (2000). Noteworthy is the relatively low developmental threshold  obtained for both life stages. 111Z. Penata Gama et al.: Dynamics of A. spiraecola in commercial alpine yarrow fields  Table 1. Estimates for the parameters of Taylor’s (1961) spatial distribution model (a,b, equation 1), for the order of the delay (k, equation 3), for the developmental rate (α, β, Tl , Tu, equation 4), and for the stage-specific intrinsic survival (λ, ξ, ς, equation 5) of Aphis spiraecola.  Life stages Spatial distribution Delay order Developmental rates Intrinsic survival a b k α β Tl Tu λ ξ ς nymphs 14.228 1.926 71 7.97E-05 0.5 2.3 35.0 0.015 0.782 0.792 adults 13 5.09E-04 0.5 2.3 35.0 Tab. 2 lists the parameter estimates for the reproductive profile, the fecundity, the  wing formation and the compound mortality function. The data provided by Wang and  Tsai (2000) were sufficient to obtain satisfactory estimates for the former two functions,  while Holst and Ruggle (1997) provided only the basic model for density-dependent wing  formation. The parameters of this function as well as the parameters of the compound  mortality function were obtained by comparing model predictions with observed  infestation patterns. Noteworthy, the compound effect depends on physiological time  rather than density. Hence, the model disregards possible density-dependent effects of  natural enemies and the host plant on A. spiraecola populations. Tab. 2. Parameter estimates for the reproductive profile (τ, υ, R, equation 6), for the temperature- dependent fecundity (ø, φ, ι, equation 7), for the density dependent wing formation (ν, ω, equation 8), and for the physiological time-dependent stage-specific compound mortality of Aphis spiraecola  (ρ, ψ, equation 9). Life stages Reproductive profile Fecundity Wing formation Compound mortality τ υ R ø φ ι ν ω ρ ψ nymphs 15.8274 112.2685 adults 7.97 1.27 42.37 0.000556 2.335 1.694 6.3677 26.4214 15.8274 112.2685 The infestation patterns of A. spiraecola in the Poschiavo yarrow field is depicted in  Fig. 3. Accordingly, A. spiraecola increases after the beginning of April until July and  decreases thereafter to low numbers. Morlacchi et al. (2010) observed a second peak in  September which is not considered in this work. In all samples, the ratio of the standard  error to the mean high was (from 0.31 to 0.94) indicating a low level of reliability of  the density estimates. Nevertheless, the low reliability is considered as sufficient for  validating the predicted infestation patterns. As expected, the disregard of wing formation and compound mortality predicts an ever  increasing population during the time under study (Fig. 3). The slow increase is due to  the relatively low temperatures of the alpine environment. The density-dependent wing  formation stabilizes the population in summer. As previously mentioned, the population  suffers in summer from losses due to natural enemies, plant senescence and cutting  Journal of Entomological and Acarological Research, Ser. II, 42 (2), 2010112 (Morlacchi et al., 2010). Only the consideration of the physiological-time dependent  compound mortality reduces the population to low levels in mid summer (Fig. 3).    Fig. 3. Simulated and observed infestation patterns of Aphis spiraecola in the Poschiavo yarrow  field (triangles for 2007 data, quadrats for 2008 data). The line indicates the simulated patterns  in absence of wing formation and compound mortalities, the dots indicate the simulated patterns  without compound effect of mortalities, the dashed line indicates the simulated pattern with wing  formation and compound mortalities (the density of 119.81 aphids on May 10, 2007, is omitted). DISCUSSION The parameters of Taylor’s power law represented in equation 1 indicate a highly  aggregated distribution among plants. The relatively big physical size of the sampling  unit (plant) may have contributed to the relatively high sampling factor of a = 14.2278.  Possibly, the selection of a whole plant sample unit rather than the consideration of plant  parts is responsible for this result (Morlacchi et al., 2010). According to the enumerative sampling plan, the optimum sample size consists of  400 – 500 plans. This indicates that the sample size of 27 plants used in this work, albeit  satisfactory for monitoring infestation patterns, is too low for estimating population  densities in analyses of the population dynamics. The optimum sample size, resulting  from the high values of the parameters of Taylor’s law (1961), requires considerable  investments in sampling activities. However, there are possibilities for reducing the efforts  even if non-destructive sampling is required. Apart from a revision of the reliability levels,  there are three opportunities for making sampling more efficient. First, it may be possible  to rely on visual examinations rather than on the beating tray technique. Second, during  the reproductive growth phase, the design of a two-stage sampling plan with plants as  113Z. Penata Gama et al.: Dynamics of A. spiraecola in commercial alpine yarrow fields  primary and stems as secondary sampling units may be feasible (e.g. Cochran, 1977).  Third, the substitution of enumerative by binomial sampling plans may also reduce the  investments into sampling activities (Morlacchi et al., 2010). In absence of wing formation and compound effects of the host plant, cutting and  natural enemy activity, the model predicts slowly increasing aphid densities (Fig. 3).  The slow increase may be due to the low temperatures in the alpine environment. At the  beginning of the infestation, higher temperatures as expected in apple and citrus growing  areas would produce a higher increase. The model parameters have been estimated from  life table data obtained on citrus with a Florida aphid biotype (Wang and Tsai, 2000).  In a subsequent paper, Tsai and Wang (2001) demonstrated how different host plants  affect the life table statistics of A. spiraecola. To put parameter estimates on a more  solid ground than done in this paper, we recommend the construction of life tables for  the alpine biotype of A. spiraecola on the Spak yarrow cultivar.   In absence of wing formation and compound effects of the host plant, cutting and  natural enemy activity, the model was parametrized with the data of Wang and Tsai  (2000) and visually compared with an independent data set. The consideration of wing  formation and compound mortality effects, however, was only possible by using the same  field data for both parameter estimation and model testing. To overcome this limitation,  more field data, preferably taken in different environments, should be collected and used  for model validation against independent data sets. Additional field and laboratory data  are also required for creating a more solid ground for the formulation of wing formation  and for extending the model towards the development of other morphs than winged and  wingless individuals, and towards the development of overwintering eggs. This would  allow a more satisfactory initialization of the model than done here. The compound mortality consists of the combined effect of natural enemies, plant  senescence and cutting (Morlacchi et al., 2010). The separation of the compound effect  into different components would undoubtedly improve the mechanistic basis of the  model. A comparison between the Poschiavo field, where the plants have been harvested,  and the Dazio field, where the grower renounced on harvesting in the year under study,  indicates that cutting strongly affects the aphid dynamics. Additional field studies on  the effects of cutting on aphid survival and the on-going studies on the interactions  between aphids and the yarrow host plant (Madeo et al., 2009), on one hand, hold the  promise for separating plant effects from predation in further model development. The  on-going studies on the interactions between aphids and natural enemies (Morlacchi et al., 2010), on the other hand, could lead to a separation of biological control from plant  effects. Brown and Matthews (2008) remind us that natural enemy activity is important  in some but not all studies on aphid population dynamics. Finally, the resulting model  could be used to study bottom-up and top-down effects in the yarrow-aphid-natural  enemy system. In a recent publication, Miller (2008) may have proposed an adequate  hypothesis by stating that it is now widely accepted that herbivore dynamics can be  influenced by both bottom-up and top-down forces, and their relative importance can  vary spatially and temporally. Journal of Entomological and Acarological Research, Ser. II, 42 (2), 2010114 ACKNOWLEDGEMENTS The daily maximum and minimum temperatures of Robbi-Poschiavo were kindly  made available by MeteoSchweiz, 8044 Zürich. REFERENCES alilla r., seVerini m., PesOlillO s., 2005 - Modello a ritardo distribuito a temperatura variabile  per la simulazione dello sviluppo ontogenetico in stadi giovanili di popolazioni peciloterme.  Rivista Italiana di Agrometeorologia, 3: 30-33.  BenDerOth M., textOr s., WinDsOr a.J., mitchell-OlDs t., GershenzO J., krOyman J., 2006  - Positive selection driving diversification in plant secondary metabolism. Proc. Natl. Acad.  Sci. USA,. 103 (24): 9118-9123. BeneDek B., GJOncaJ n., saukel J., kOPP B., 2007 - Distribution of phenolic compounds in  Middleeuropean Taxa of the Achillea millefolium L. Aggregate. Chem. Biodivers., 4:849-857. Bianchi G., BaumGärtner J., Delucchi V., rahaliVaVOlOlOna n., 1990 - Modèle de population  pour la dynamique de Maliarpha separatella Ragonot (Pyralidae, Phycitinae) dans les rizières  du lac Alaotra. J. Appl. Ent., 110: 384-397. BrOWn m.W., 2004 - Role of aphid predator guild in controlling Spiraea aphid populations on  apple in West Virginia, USA. Biological Control, 29: 189–198. BrOWn m.W., matheWs c.r., 2008 - Conservation biological control of spirea aphid, Aphis spiraecola (Hemiptera: Aphididae) on apple by providing natural alternative food resources.  Eur. J. Entomol., 105: 537-540. Brière J.-F., PracrOs P., le rOux a.y., Pierre J.s., 1999 - A novel rate model of temperature- dependent development for arthropods. Environ. Entomol., 28: 22-29. cOchran W.G., 1977 - Sampling Techniques. 3rd ed., Wiley, New York, NY, USA. curry G.L., FelDman R.M., 1987 - Mathematical Foundations of Population Dynamics. Texas  Engineering Experiment Station (TEES). The Texas A&M University System, Texas A&M  University Press, College Station, TX, USA, 246 p. Di cOla G., GiliOli G., BaumGärtner J., 1999 - Mathematical models for age-structured  population dynamics. - Ecological Entomology. huFFaker c.B., Gutierrez a.P. (eds.), 2nd  edition, Wiley, New York, pp.503-534. Fraenkel G.s., 1959 - The raison d’être of secondary plant substances. These odd chemicals  arose as a means of protecting plants from insects and now guide insects to food. Science,  129: 1466-1470.  Gutierrez A.P., 1996 - Applied Population Ecology. A Supply-Demand Approach. Wiley, New  York, NY, 300 p. Gutierrez A.P., BaumGärtner J., 2007 - Modeling the dynamics of tritrophic population  interactions. - Perspectives in Ecological Theory and Integrated Pest Management. Kogan  M., Jepson P. (eds.), Cambridge University Press, Cambridge, UK, pp. 301-360. hermOsO De menDOza a., arOuni r., BelliOre B., carBOnell e.a., Pérez-PanaDés J., 2006 -  Intervention thresholds for Aphis spiraecola (Hemiptera: Aphididae) on Citrus clementina.  J. Econ. Entomol., 99(4): 1273-1279. hOlst, n., ruGGle P., 1997 - A physiologically based model of pest–natural enemy interactions.  Exp. Appl. Acarol., 21: 325–341. 115Z. Penata Gama et al.: Dynamics of A. spiraecola in commercial alpine yarrow fields  hOnG J., han k.s., BOO k.s., 2003 - Sex pheromone of Aphis spiraecola (Homoptera: Aphididae):  composition and circadian rhythm in release. J. Asia Pacific Entomol., 6: 159-165. hOy m.a., ru n., 2008 - Brown Citrus Aphid Parasitoid, Lipolexis scutellaris Mackauer (Insecta:  Hymenoptera: Aphidiidae). Document EENY-181 (IN338) of the Featured Creatures series  of the Entomology and Nematology Department, Cooperative Extension Service, Institute of  Food and Agricultural Sciences, University of Florida. Published: November 2000. Reviewed:  March 2008. Available online [http://edis.ifas.ufl.edu/pdffiles/IN/IN33800.pdf] [retrieved on  May 19, 2010]. hutchisOn W.D., hOGG D.B., POsWal m.a., BerBeret m.c., cuPerus G.W., 1988 - Implications  of the stochastic nature of Kuno’s and Green’s fixed-precision stop lines: sampling plans for the  pea aphid (Homoptera, Aphididae) in alfalfa as an example. J. Econ. Entomol., 81: 749-758. karanDinOs m.G., 1976 - Optimum sample size and comments on some published formula. Bull.  Entomol. Soc. Amer., 22: 417-421. limOnta l., mOrlacchi P., BaumGärtner J., 2009a - Modello stocastico-biofisico per lo sviluppo  di una coorte di Oulema duftschmidi a temperature costanti - Modelli per la difesa delle piante.  Monografie e articoli scientifici presentate alle quarte giornate di studio su metodi numerici,  statistici e informatici nella difesa delle colture agrarie e forestali: ricerca ed applicazione.  seVerini m., PesOlillO, s. (eds.), Viterbo, marzo 2007. Aracne, Roma: 177-181. limOnta l., lOcatelli D., BrOGlia t., BaumGärtner J., 2009b - Cohort developmental models  for integrated Corcya cephalonica (Stainton) population management. Boll. Zool. Agr.  Bachic., 41: 215-226. maDeO m., minGOzzi m., mOrlacchi P., GiOrGi a., Onelli e., lOzzia G.c., BaumGärtner J., 2009 - Aphid infestations influence growth and bioactive compounds, and induce resistance  in Achillea collina Becker ex Rchb. - Oral presentation at: Plants from High Altitude,  Phytochemistry and Bioactivity. A joint Symposium of the Phytochemical Society of Europe  (PSE) and the Austrian Pharmaceutical Society (OPhG). 26-29 April 2009. University Centre  Obergurgl, Tyrol, Austria. miller t.e.x., 2008 - Bottom-up, top-down, and within-trophic level pressures on a cactus-feeding  insect. Ecol. Entomol., 33: 261-268. mOrlacchi P., Brunella m., GiOrGi a., lOzzia G.c., BaumGärtner J., 2010 - Yarrow (Achillea collina) yield formation, aphid population fluctuations and sampling plans (Macrosphoniella millefolli, Aphis spiraecola, Coloradoa achilleae) in commercial alpine fields. J. Entomol.  (subm. 2010). Rykiel E.J. jr., 1996 - Testing ecological models: the meaning of validation. Ecol. Model., 90:  229-244.  seVerini m., 2009 - Base matematica del Modello a Ritardo Variabile (Time Invariant Delay)  - Modelli per la difesa delle piante. Monografie e articoli scientifici presentate alle quarte  giornate di studio su metodi numerici, statistici e informatici nella difesa delle colture agrarie  e forestali: ricerca ed applicazione. seVerini m. PesOlillO s. (eds.), Viterbo, marzo 2007.  Aracne, Rome, pp. 13-57. taylOr l.r., 1961- Aggregation, variance and the mean. Nature. 189:731-735. tsai J.h., WanG J.-J., 2001 - Effects of host plants on biology and life table parameters of  Aphis spiraecola (Homoptera: Aphididae). Environ. Entomol., 30: 44-50. Vansickle J., 1977 - Attrition in distributed delay models. IEEE Trans. Syst. Man Cybern., 7:  635-638. WanG J.-J., tsai J.h., 2000 - Effect of temperature on the biology of Aphis spiraecola (Homoptera:  Aphididae). Ann. Entomol. Soc. Am., 93: 874-883. Journal of Entomological and Acarological Research, Ser. II, 42 (2), 2010116 Wink m., 2003 - Evolution of secondary metabolites from an ecological and molecular phylogenetic  perspective. Phytochemistry, 64: 3-19. zulFaiDah Penata Gama, M.Si., Biology Department, MIPA Faculty, Brawijaya University, Jl.  Veteran, Malang 65145, Indonesia. PaBlO mOrlacchi, GiusePPe carlO lOzzia, JOhann BaumGärtner, Dipartimento di Protezione  dei Sistemi Agroalimentare e Urbano e Valorizzazione delle Biodiversità (D.I.P.S.A.),  Università degli Studi di Milano, Via Celoria 2, 20133 Milano, Italy. Anna GiOrGi, Dipartimento di Produzione Vegetale (Di.Pro.Ve.), Università degli Studi di Milano,  Via Celoria 2, 20133 Milano, Italy. Accepted 27 August 2010