Validating spatiotemporal predictions of western corn rootworm at the regional scale (Tuscany, central Italy)Susanna Marchi1*, Diego Guidotti2, Massimo Ricciolini3, Ruggero Petacchi1

Ital

ian

Jour

nal o

f Agr

omet

eoro

logy

- 3/

2017

Riv

ista

Ita

liana

di A

grom

eteo

rolo

gia

- 3/2

017

13

Abstract: New invasive pest species are difficult to manage, because researchers, advisors, and farmers in the newly invaded areas lack advanced understanding in how and when managing them. The aim of this study was to test selected models that could reliably predict the phenology of Diabrotica virgifera virgifera (LeConte), an important insect pest of maize. We compared the results of three years monitoring activity in three areas of Tuscany with the output of two different models, in order to predict adult emergence from air temperature measurements. The best results were achieved with the model that utilized the date of maize planting as starting date for the accumulation of degree-days, confirming a strict connection between crop and pest phenology. Model output for the predicted day of the year for start and peak of the pest cumulative emergence was mapped over the administrative boundaries of Tuscany with a regression model run with temperatures derived from WorldClim on-line database. These results will be integrated in a Decision Support System for containment and management strategies of maize pests in Tuscany.Keywords: Diabrotica virgifera virgifera, pest monitoring network, spatial forecasting models.

Riassunto: Le nuove specie invasive di insetti nocivi sono difficili da controllare, perché ricercatori, tecnici, e agricoltori nelle nuove aree oggetto di invasione mancano degli strumenti necessari per stabilire come e quando trattare. Lo scopo del presente lavoro è stato di provare modelli che possano predire in maniera affidabile la fenologia di Diabrotica virgifera virgifera (LeConte), un importante insetto del mais. Abbiamo confrontato i risultati di tre anni di monitoraggio in tre aree della Toscana con i risultati di due modelli, per predire l’emergenza degli adulti in base alla temperatura dell’aria. I risultati migliori sono stati ottenuti con il modello che utilizza la data di semina come inizio per l’accumulo dei gradi giorno, confermando la stretta connessione fra la fenologia dell’insetto e della coltura. L’uscita del modello per la previsione del giorno dell’anno d’inizio e picco d’emergenza cumulata è stato mappato sulla Toscana con un modello regressivo alimentato con le temperature derivate dal database WorldClim disponibile on line. Questi risultati saranno integrati in un Sistema di Supporto alle Decisioni per la coltivazione sostenibile del mais in Toscana.Parole chiave: Diabrotica virgifera virgifera, reti di monitoraggio di insetti, modelli spaziali predittivi.

* Corresponding author’s e-mail: susanna.marchi@santannapisa.it1 Scuola Superiore Sant’Anna, Institute of Life Science, Pisa. 2 AEDIT srl, Pontedera. 3 Servizio Fitosanitario - Regione Toscana, Firenze.Submitted 02 September 2016, accepted 25 January 2017

DOI:10.19199/2017.3.2038-5625.013

IntRoductIonThe western corn rootworm (WCR), Diabrotica virgifera virgifera, LeConte (Coleoptera: Chry-somelidae), is one of the most serious insect pests in the USA for maize (Zea mays L.) (Miller et al., 2005; Meinke et al., 2009; Spencer et al., 2009). WCR is univoltine and overwinters as egg normally in the soil of the maize field. The larvae hatch during the spring and feed almost exclusively on maize roots (Oyediran et al., 2004). Adults emergence begins in late June early July and can extend through the beginning of September (Spencer et al., 2009). WCR adults feed on various vegetative parts of cultivated maize, while larvae feed on maize roots, causing damage, eventually leading to stalk lodging. The feeding by larvae weakens the root system of maize plants, reducing photosynthesis (Godfrey et

al., 1993; Riedell and Reese, 1999; Urias-Lopez et al., 2001), as well as nitrogen and water uptake, making the crop more susceptible to drought and disease (Estes et al., 2016). Accidental and multiple introductions in Europe occurred between the late 1980s and early 2000s (Ciosi et al., 2008; Szalai et al., 2011). Nowadays, WCR is regarded as a serious threat to European maize production, causing yield reductions and costly chemical applications, especially in areas with high crop densities; yield losses of 10-30% are reported in south eastern Europe, 20-30% in Italy (Kehlenbeck and Krügener, 2014). Traditional control methods include crop rotation, which has been a widely used control strategy in the USA Corn Belt (Spencer et al., 2009) and Europe (Boriani et al., 2006). However, when the maize/soybean crop rotation management strategy became widespread in parts of the eastern Corn Belt (USA), it selected for a WCR population, which turned behaviourally resistant to crop rotation (Levine and Oloumi-Sadeghi, 1996). Rotation-resistant

ciano 3-17.indb 13 24/11/17 11.49

14

Ital

ian

Jour

nal o

f Agr

omet

eoro

logy

- 3/

2017

Riv

ista

Ita

liana

di A

grom

eteo

rolo

gia

- 3/2

017

are very different from those in northern Italy, in terms of average field size, crop rotation practises and pesticide use and regulation, and require specific understanding of environmental factors and their spatial interactions to direct pest management decisions at the regional scale. Tuscany represents also a good test for model validation at the regional level, its landscape being a complex mosaic of varying agroclimatic patches. The use of spatial statistics and phenological models for predicting geographic distribution and temporal dynamic of insect populations is widely used for increasing the efficacy of area wide pest management (AWPM) programs (e.g., Liebhold et al., 1993; Petacchi et al., 2015). In particular, the development of GIS-based (Geographic In- formation System) DSS (Decision Support System) has brought many advantages in combining territorial and meteorological information. Several studies have addressed the problem on how to use temperature-driven simulation models in pest management (e.g., Régnière, 1996). Degree-day prediction of insect populations is a tool widely used for timing management decisions and the prophylactic use of insecticides. In USA, an AWPM initiative was implemented in 1996 to combat WCR (Chandler and Faust, 1998; Beckler et al., 2004). In Europe, early predictions of distribution and abundance of adult WCR populations over expanding area of influence could help appropriate timing for pesticide applications, diminishing their amount, as well as support adapting strategies to warming-induced shifts of pest outbreaks and changes. The modelling approach is an important instrument in phytosanitary risk assessment at the large scale (Baker, 1991), which allows to better understand and predict the dynamics of insect populations under changing environmental conditions and management practices. Several models have been implemented to forecast and define WCR phenology and to provide support to farmers in selecting the correct time for spraying. These are based on the assumption that insects are poikilotherms: temperature is the major abiotic factor influencing their life cycle. Indeed, insect development occurs within a specific temperature range, known as the ‘thermal range’, with minimum and maximum thresholds (Dixon et al., 2009). These models are either based on the calendar date, air or soil degree-day accumulations, with specific development thresholds (Ruppel et al., 1978; Bergman and Turpin, 1986; Fisher et al., 1990; Davis et al., 1996; Nowatzki et al., 2002; Stevenson et al., 2008).

populations may lay eggs outside maize fields, in alfalfa, winter wheat, rye, or soybean (Levine et al., 2002), and also feed on roots of other grasses besides corn (Rondon and Gray, 2003; Clark and Hibbard, 2004; Schroeder et al., 2005). However in Italy and in Europe, to date, the approach of crop rotation is working relatively well in WCR check (Boriani et al., 2006; Furlan et al., 2006). Moreover, crop rotation is beneficial to the environment and the crop (water quality, soil fertility, etc.) and should be maintained as control strategy. On non-rotated maize, the most common control strategies are soil insecticides applied at planting to control larvae and aerial applications in summer against adults. However, the indiscriminate use of soil insecticides has promoted environmental contaminations and economic concerns (Gray et al., 1993). The early measures to prevent the spread of WCR within the EU (Decision 2003/766/EC) did not prove to be successful, which brought to their further integration (Directive 2014/19/EU and Decision 2014/62/EU). The partial efficacy of agronomic practices (including transgenic hybrids) and the development of insecticide resistance, observed in the USA, have prompted the need for new management strategies (Tollefson, 1998; Levine et al., 2002; Gassmann et al., 2011). Nowadays, European countries are called to envisage effective monitoring and sustainable control of WCR (Recommendation 2014/63/EU), considering that this pest has been withdrawn from harmful organism with quarantine status, as it is neither feasible to pursue its eradication from the EU, nor to prevent its further spread into the areas that are currently free. WCR is currently spreading throughout Europe (Moeser and Guillemaud, 2009), including north and central Italy. In Tuscany, the first scouting was conducted in 2010, restricted to the northeast area of the region, near Emilia Romagna, where WCR was noticed few years earlier. Throughout the first monitoring campaign, WCR was caught northwest of Florence (Vicchio). Since its first capture, the number of WCR individuals caught has gradually increased in the following years. Maize is one of the major crops in Tuscany, and its production is important both to regional economy and to rural communities due to the increased share of renewable energy sources. The number of anaerobic digester plants have boosted in the last decade in agricultural lands suitable for maize cultivation, which has resulted in a larger surface area used for growing biogas maize. Maize is also cultivated for maize grain to be used for livestock feeding. Maize agricultural conditions in Tuscany

ciano 3-17.indb 14 24/11/17 11.49

Ital

ian

Jour

nal o

f Agr

omet

eoro

logy

- 3/

2017

Riv

ista

Ita

liana

di A

grom

eteo

rolo

gia

- 3/2

017

15

- 43°46’19.6” N, 10°25’32.6” E), south (area 3 - Grosseto province - 42°49’03.8” N, 11°02’48.3” E) (Fig. 1). In the nine experimental fi elds (three per area), the hybrids planted were FAO 500-700 rating; 120-140 days total time. The nine fi elds, although with different size, were selected in areas where maize was the dominant crop. Conventional tillage methods were used, with no insecticides at planting. Other agronomic practises, like fertilization and irrigation, were at the discretion of the grower and typical of the maize production system in Tuscany. Planting date varied rather widely according to weather conditions and grower decisions, from mid-March to early June. The harvest time varied from August until October/November, depending on weather conditions, hybrid growing period and the destination of the production (grain, seed, silage or energy). Seed treatments are often used to control pests that attack maize roots. Insecticide treatments to control European corn borer, Ostrinia nubilalis Hbn., were not applied.Tuscany is characterized by typical Mediterranean climate, with hot and dry summers and mild winters. However, its territory is portrayed by a

Davis et al. (1996) developed a phenology model based on air temperature. These authors obtained the best fi tting either using minimum developmental thresholds near 6 °C for immature stages and 1-2 °C for adult emergence, or calculating degree-days setting a minimum threshold of 11 °C and a maximum threshold of 18 °C. Nowatzky et al. (2002) predicted adult emergence, beginning from a biofx defi ned as the date of fi rst beetle emergence in a fi eld. Cumulated air temperature degree-days explained 85% of the variability in total WCR emergence over 5 years. This model explained 89% and 83% of the variability in total beetle emergence observed in the validation year, depending on the monitoring approach. Stevenson et al. (2008) developed and validated a degree-day phenology model to predict the emergence of WCR adults. These authors determined the functional lower development (base) temperature, the optimum starting date, and the sum of degree-days for phenological events from onset to 99% adult emergence.The adoption of models that reliably predict WCR emergence would allow scouting to be focused on key periods, such us peak of emergence, rather than over the entire season. Accurately predicting the fl ight period may also help matching WCR behavioural patterns and population control measures at the regional level. The objective of this study was to test models implemented in other countries/environments that can reliably predict adult WCR emergence in Tuscany, so that scouting and eventual adult insecticide applications can be accurately timed. We compared the results of three years monitoring activity in three areas of Tuscany with previously published models for estimating their validity in multifaceted landscapes. We hypothesized a better tuning of adult emergence predictions at the regional level by comparing different models that forecast insect fl ights from air temperatures. We focused on two straightforward models based on temperature, but with different starting day (biofi x) for degree-day accumulation, which may be relevant for their application potential in GIS-based DSS for AWPM.

MAteRIAls And Methods

study area This study was carried out in Tuscany, north-central Italy, from June to September 2013–2015. The monitored fi elds were located in three different areas suitable for maize production: north (area 1 - Massa province – 44°19’37.1” N, 9°53’34.2” E), north-centre (area 2 - Pisa and Lucca provinces

Fig. 1 - Map of the distribution of the nine experimental sites in the three different areas of Tuscany. Area 1 = Massa Carrara province, Area 2 = Pisa and Lucca provinces, Area 3 =Grosseto province. The DEM source is the ASTER Global Digital Elevation Model (GDEM) v2 data with a resolution of 30 m.Fig. 1 - Mappa della distribuzione dei 9 punti di monito-raggio nelle tre aree della Toscana. Area 1 = provincia di Massa Carrara, Area 2 = provincie di Pisa e Lucca, Area 3 = provincia di Grosseto. Il DEM utilizzato è ASTER Global Digital Elevation Model (GDEM) v2 data con una risoluzi-one di 30 m.

ciano 3-17.indb 15 24/11/17 11.49

the beginning of adult fl ight, for Novatzki-2 the biofi x is the day in which the fi rst adult has been detected in the fi eld by means of pheromone traps. 2. Stevenson et al. (2008), in which the biofi x is the maize planting date.For each area and year, we calculated emergence expressed as cumulative percentage of the total and we compared the output of the models for 10%, 50% and 90% of the cumulated emergence. To evaluate the predictive accuracy of the models, on the basis of the lowest RMSE (root mean square error) and MAE (mean absolute error), we compared the deviation in emergence in actual days, between the current data set (i.e., observed) and those predicted through modelling.

data spatializationStevenson model (Stevenson et al., 2008) was run on about 110 weather stations of the regional network for 2013, 2014 and 2015. As starting date of the model, we considered the average planting date as obtained from the regional monitoring network (200 points in 2013, 80 in 2014 and 80 in 2015), between 29 April and 6 May, depending on the year. The prediction of the day of the year (DOY) for the beginning (10%) and peak (50%) of the cumulated adult emergence was calculated for the three years. Spatial fi t of the model was carried out with a regression model by using interpolated baseline climate data obtained from WorldClim database (Hijmans et al., 2005; http://www.worldclim.org/). We created a raster fi le averaging the monthly mean temperatures for Tuscany from March to August (from maize planting to harvest). We assigned to each weather station the corresponding value of WorldClim raster. DOY calculated for each weather station, by means of the Stevenson model, for 10% and 50% of the adult emergence, was correlated with the log-transformed WorldClim derived average monthly mean temperature data, generating the regression model:

DOY = intercept + slope • ln(WC_March-August) where DOY is the day of the year for the 10% or 50% of cumulated adult emergence; intercept and slope are the regression parameters; WC_March-August is the average of March-August monthly temperature from WorldClim (expressed in tenths of a degree).

By the use of WorldClim data, the geographical factors, as derived by the topography of the rough territory, which can affect the weather and result in microclimatic patchiness and strong changes (temperature) over short distances, were taken into account.

16

Ital

ian

Jour

nal o

f Agr

omet

eoro

logy

- 3/

2017

Riv

ista

Ita

liana

di A

grom

eteo

rolo

gia

- 3/2

017

marked climatic variability due to the geographic position and orographic transformation, namely the alternation of imposing massifs, hills and plains. The Apennine chain is opposed to the transfer of air masses of north-eastern origin. Whereas, the Apuan Alps and the coastal hills reduce the effect of westerlies. The climate shows, therefore, sharp contrasts; in the face of a mild coastal climate, conditions resembling those of continental areas may occur inland. Regional annual mean temperature in about 14 °C. Annual average precipitation varies from a minimum of 600 mm in the central and southern coastal area to 2000-3000 mm in the Apuan Alps (Niccolai and Marchi, 2005).There was one agrometeorological station of the regional monitoring network (Tuscany Region), at each site, close to one of the monitored fi elds. The distance from the agrometeorological station and the monitored fi elds varied from 0 (in-fi eld station) to 15 km. Daily minimum and maximum temperature data for model testing were obtained from these stations.

Monitoring activityThe fi elds were selected in the three areas in which the highest number of individuals was counted the previous year (2012, the fi rst year of monitoring activity at the regional scale conducted by the Tuscany Phytosanitary Service). Adult fl ights were monitored by means of pheromonic traps (Diabrotica Track, SERBIOS s.r.l., Badia Polesine, Italy). From mid-June (2013, 2014 and 2015), two traps were placed in each of the three fi elds per area. The 18 traps were monitored weekly, counting and removing all the individuals; traps and pheromone were replaced with fresh ones every 20 days. Flight trend was described for each area by calculating the number of adult catches per trap in each year.

Model testingWe tested two models based on a theoretical start date, and on lower and upper functional development temperatures over the given period (which were equivalent to the minimum and maximum temperature thresholds). Air temperature degree-days, starting from the different date (biofi x) for each model and for each fi eld, were calculated with the sine wave method (Allen, 1976). Calculation parameters for each model are the following:1. Novaztki et al. (2002) with two biofix; for Novatzki-1 the accumulation started with the output day of Davis model (Davis et al., 1996) for

ciano 3-17.indb 16 24/11/17 11.49

17

Ital

ian

Jour

nal o

f Agr

omet

eoro

logy

- 3/

2017

Riv

ista

Ita

liana

di A

grom

eteo

rolo

gia

- 3/2

017

Once built the model, we calculated the estimated DOY for each cell of the WorldClim raster.

Results

environmental conditionsTrends of daily minimum and maximum temperatures, and precipitation from January to August in the three areas for the three years are shown in Fig. 2. In 2013, average daily maximum and minimum temperatures of, respectively, 18.4 and 9.1 °C in area 1, 20.8 and 10.5 °C in area 2, and 21.6 and 9 °C in area 3 were recorded. Cumulated precipitation was 1273, 782.6 and 530.6 mm in area 1, 2 and 3, respectively. In 2014, average maximum and minimum daily temperatures from January to August were, respectively, 19.4 and 10.1 °C in area 1, 20.8 and 10.4 °C in area 2, and 21.7 and 9.8 °C in area 3. Total annual precipitation was more abundant than in 2013: 1466.8 in area 1, 943.6 in area 2, and 680.2 in area 3. In 2015, we recorded the lowest precipitation: 520.6 mm in area 1, 557.4 mm in area 2, and 412.8 mm in area 3. Average minimum and maximum daily temperatures were, respectively, 20.5 and 10.4 °C

in area 1, 21.6 and 10.3 in area 2, and 22.8 and 10 °C in area 3.

Monitoring campaignsIn 2013, adult emergence started on 1 July (DOY 182) in area 2, on 6 July (DOY 187) in area 1 (Fig.3), with no captures in area 3. In 2014, adult catches began on 19 June (DOY 171) in area 2, 4 July in area 1 (DOY 186), and 9 July (DOY 191) in area 3. In 2015, adult emergence started earlier: 22 June (DOY 173) in area 3, 24 June (DOY 175) in area 2, and 4 July (DOY 185) in area 1. Emergence was completed in all fields by late August or early September. In 2013, a total of 684 WCR beetles were captured in area 1, 233 in area 2, and 0 in area 3 (Fig. 3). In 2014, catches were 320 in area 1, 285 in area 2, and 25 in area 3. In 2015, catches were 195 in area 1 (two fields were in rotation), 482 in area 2, and 171 in area 3. In 2013, two flight peaks were observed in area 1, one on DOY 201 (20 July) and the second on DOY 220 (8 August) (Fig.3); while in area 2, the peak was reached on DOY 217 (5 August). In 2014, close peaks were recorded in the three monitored

Fig. 2 - Daily maximum and minimum average temperatures and rainfall in the three experimental areas from January to August during 2013, 2014 and 2015. Parameters and years are referred to by the legend in the figure. Area 1 = Massa Carrara province, Area 2 = Pisa and Lucca provinces, Area 3 =Grosseto province.Fig. 2 - Temperatura massima e minima e precipitazioni giornaliere nelle tre aree sperimentali da Gennaio ad Agosto del 2013, 2014 e 2015. Parametri ed anni sono indicati in figura. Area 1 = provincia di Massa Carrara, Area 2 = provincie di Pisa e Lucca, Area 3 = provincia di Grosseto.

ciano 3-17.indb 17 24/11/17 11.49

18

Ital

ian

Jour

nal o

f Agr

omet

eoro

logy

- 3/

2017

Riv

ista

Ita

liana

di A

grom

eteo

rolo

gia

- 3/2

017

when no beetles were captured, and the date when the first beetle of either sex was observed in a trap.In area 1 (northern Tuscany), adult emergence reached 10% of the cumulative total on 15 July (DOY 196) in 2013, 11 July (DOY 193) in 2014, and 13 July (DOY 195) in 2015. 50% of adult emergence was observed on 28 July (DOY 209) in 2013, 20 July (DOY 201) in 2014, and 23 July (DOY 204) in 2015. The end of adult emergence (90% of the total) was recorded on 18 August (DOY 230) in 2013, 11 August (DOY 223) in 2014, and 2 August (DOY 215) in 2015.In area 2 (central Tuscany), 10% of adult emergence was recorded on 16 July (DOY 197) in 2013, 4 July (DOY 185) in 2014, and 7 July (DOY 188) in 2015. 50% of adult emergence was observed on 4 August (DOY 216) in 2013, 24 July (DOY 205) in 2014, and 24 July (DOY 205) in 2015. The end of adult emergence (90% of the total) was recorded on 26 August (DOY 238) in 2013, 22 August (DOY 234) in 2014, and 11 August (DOY 224) in 2015.Finally, in area 3 (southern Tuscany), no catches were recorded in 2013. In 2014, catches were only in one field: 10% on 13 July (DOY 194), 50% on 25 July (DOY 206), and 90% on 23 August (DOY 235). In 2015, cumulative catches were 10% on 3 July (DOY 184), 50% on 12 July (DOY 193), and 90% on 23 July (DOY 204).The best predictive accuracy (predicted emergence) was obtained with Stevenson and Nowatzky-2 models with biofix at the first adult captured in the field. Nowatzky-1 with biofix at the emergence date following Davis model gave a poor prediction for 10%, 50%, and 90% of the cumulated catches. For the beginning of the emergence, 10%, the best model was Nowatzky-2, with a difference between predicted and observed value of 1-5 days and a RMSE between 1 and 7. Stevenson model predicted the starting of emergence (10% of the total cumulated) with a difference between predicted and observed values between 1 and 12 days (worst result in 2014) and a RMSE from 3 to 14. The 50% of the cumulative emergence (50%) was well predicted by Stevenson model in terms of number of days; this model showed a difference between predicted and observed day of 0 to 7 days with RMSE of 3 to 15. Nowatzki-2 model showed difference between predicted and observed of 1-11 days and RMSE of 3-13. The end of emergence (90% of the cumulated catches) was predicted by Nowatzki-2 and Stevenson models with few days of difference. Nowatzki-2 model showed a difference in days between 1 and 17 with RMSE of 2-17. Stevenson model showed a difference in days between 2 and 11 and RMSE of 4-22 (Tab. 1).

areas: DOY 206 (25 July) in area 1, DOY 209 (28 July) in area 2, and DOY 208 (27 July) in area 3. In 2015, the peak emergence occurred on DOY 207 (26 July) in area 1, DOY 202 (21 July) in area 2, and DOY 193 (12 July) in area 3.

Model comparisonAccording to Nowatzki et al. (2002), the day for the beginning of emergence (starting of adult flight) was set as the date midway between the last date,

Fig. 3 - Number of adults captured in pheromone traps from June to August in the three fields of the three experimental areas in 2013, 2014 and 2015. For each day of the year (DOY) and for each area, data are the mean of three observations ± standard error (n=3). Area 1 = Massa Carrara province, Area 2 = Pisa and Lucca provinces, Area 3 =Grosseto province.Fig. 3 - Numero di adulti catturati nelle trappole a feromone nei tre campi delle tre aree di monitoraggio nel 2013, 2014 e 2015. Per ogni giorno dell’anno (DOY) e per ogni area i valori sono la media di tre osservazioni ± l’errore standard (n=3). Area 1 = provincia di Massa Carrara, Area 2 = provincie di Pisa e Lucca, Area 3 = provincia di Grosseto.

ciano 3-17.indb 18 24/11/17 11.49

19

Ital

ian

Jour

nal o

f Agr

omet

eoro

logy

- 3/

2017

Riv

ista

Ita

liana

di A

grom

eteo

rolo

gia

- 3/2

017

particularly for the anticipated or delayed adult emergence (Fig. 4).

dIscussIonThe spatialization of Stevenson model trough temperature data from WorldClim provided a valuable solution for implementing a DSS for AWPM. This straightforward approach would reduce the cost and effort of monitoring networks at

spatial fittingWe selected Stevenson model for spatialization purposes, since providing comparable results to Nowatzky-2 with less input requirements. The regression model between DOY of Stevenson model for 10% and 50% of cumulated adult emergence and mean temperature derived from WorldClim resulted in very good agreement (Tab. 2). The territorial variability was well described,

NOWATZKI-1 NOWATZKI-2 STEVENSONPercentage

of emergence P-o MAe RMse P-o MAe RMse P-o MAe RMse

AReA 1

201310% -18 18 18 -3 3 3 7 7 7

50% -17 17 17 -3 3 3 7 7 790% -15 15 16 -1 2 2 10 10 10

201410% -29 29 29 1 1 1 8 8 10

50% -23 23 23 5 5 6 -5 9 1090% -19 19 20 -1 4 5 10 22 22

201510% -31 31 31 1 4 4 -6 6 6

50% -27 27 27 3 4 5 -4 4 490% -15 15 15 16 16 17 3 6 6

AReA 2

201310% -38 38 39 -4 4 5 8 10 12

50% -44 44 44 -11 11 13 0 3 390% -42 42 42 -9 12 13 3 3 5

201410% -38 38 39 -3 3 4 12 12 14

50% -42 42 45 -9 10 13 5 14 1590% -47 47 48 -15 15 16 6 16 17

201510% -33 33 34 -5 6 7 6 6 8

50% -37 37 38 -10 10 12 0 6 690% -32 32 33 -6 6 7 2 4 4

AReA 3

201410% -42 42 1 1 -3 3

50% -40 40 1 1 -1 190% -43 43 -4 4 -6 6

201510% -29 29 29 2 2 2 1 3 3

50% -25 25 25 5 5 6 2 2 390% -12 12 13 17 17 17 11 11 11

tab. 1 - Mean difference between day of the year predicted by Nowatzki-1 model (biofix at the day of estimated emergence by Davis et al. (1996) model), Nowatzki-2 model (biofix at the day of first adult captured in field), and Stevenson model (biofix at planting date) (P) and day of the year observed (O) for the 10%, 50% and 90% of cumulative emergences in the three experimental areas for each year of observation. RMSE (root mean square error) and MAE (mean absolute error) are reported. Positive number of days indicates emergence date predicted by model occurred later than observed, whereas negative numbers indicate that it occurred earlier than observed. Area 1 = Massa Carrara province, Area 2 = Pisa and Lucca provinces, Area 3 = Grosseto province.Tab. 1 - Scarto tra il giorno dell’anno predetto dal modello Nowatzki-1 (inizio dell’accumulo alla data simulata per l’emergenza dal modello Davis et al., (1996)), dal modello Nowatzki-2 (inizio dell’accumulo nel giorno del primo adulto catturato in campo), e dal modello Stevenson (inizio dell’accumulo alla data di semina) (P) ed il giorno dell’anno osservato (O) per il 10%, 50% e 90% delle catture totali cumulate per ciascuna area e per ciascun anno di osservazione. RMSE (scarto quadratico medio) e MAE (errore assoluto medio) sono inoltre riportati. Numero positivo indica che il giorno previsto dal modello è pos-teriore a quello osservato, mentre numero negativo indica che il giorno previsto dal modello è antecedente a quello osservato. Area 1 = provincia di Massa Carrara, Area 2 = provincie di Pisa e Lucca, Area 3 = provincia di Grosseto.

ciano 3-17.indb 19 24/11/17 11.49

the regional level, by focusing on target areas and the most critical period of the season. Due to the large variability in environmental conditions and their interaction with management options, designing a protocol for regional monitoring requires a selection of samples in the most significant zones, thus increasing the chance of successful application of AWPM. Beckler et al. (2004) concluded that an understanding of the interactions between WCR population dynamics and environmental variables provides straightforward information to pest managers, and this can be used to identify patterns in the landscape that promote high insect population density patches to improve pest management strategies. The information derived from spatial analyses and the utilization of climatic data (Marchi et al., 2016) can help planning detailed surveys in AWPM, predicting WCR populations and formulating informed decisions, while using limited funds and human resources. Indeed, in risk analysis, resource managers need accurate maps of species distribution and abundance (Kumar et al., 2014). Economic damage caused by WCR has been observed in bordering regions of northern Italy since 2002 (Boriani et al., 2006). In Tuscany, the monitored populations were not large enough to cause economic damage and no lodging plants showing symptoms of feeding larvae were observed during the monitoring activities. In addition, no adults feeding on silks or interferences with pollination were detected. However, the growth in demand for bioenergy production is leading to larger areas being planted with maize continuously over long periods, which may increase the potential

yield losses in monoculture of maize due to WCR outbreaks. Therefore, mapping the inter-annual expansion of distribution of this pest can become effective tool to anticipate its control.The maps developed in this study can be refined at the regional scale by integrating detailed pest monitoring schemes for early warning and using scientifically designed field surveys for detailed data, as well as high-resolution predicting variables (Petacchi et al., 2015). These maps can also serve to test alternative hypotheses for WCR occurrence in non-infested districts. Relevant results will be useful

20

Ital

ian

Jour

nal o

f Agr

omet

eoro

logy

- 3/

2017

Riv

ista

Ita

liana

di A

grom

eteo

rolo

gia

- 3/2

017

Percentage of emergence year slope intercept R2

10% 2013 -72.26 574.57 0.72910% 2014 -111.2 768.38 0.74910% 2015 -54.4 469.01 0.66150% 2013 -101.5 736.28 0.74350% 2014 -150.3 983.36 0.76350% 2015 -68.06 550.25 0.618

tab. 2 - Results of the regression model (P<0.05) between day of the year (DOY) of the start (10%) and the peak (50%) of cumulated adult emergence calculated with Stevenson model and the derived WorldClim average monthly tem-perature value from March to August (see Data Spatializa-tion paragraph for the function).Tab. 2 - Risultati del modello regressivo (P<0.05) tra il giorno dell’anno per l’inizio (10%) ed il picco (50%) dei voli per gli anni 2013, 2014 e 2015 calcolato con il modello Ste-venson e la media delle temperature mensili da Marzo ad Agosto derivate da WorlClim (vedi paragrafo Data Spatial-ization per la funzione).

Fig. 4 - Distribution of day of year (DOY) of 10% (left) and 50% (right) of the cumulative emergence across Tus-cany for 2013, 2014 and 2015 using the regression model (DOY predicted by Stevenson model and the WorldClim average temperature from March to August of each year; see Data Spatialization paragraph for the funcion). Ranges of expected DOY is indicated by the colour scale in the legend. Fig. 4 - Distribuzione del giorno dell’anno (DOY) previsto per il 10% ed il 50% dell’emergenza cumulata sulla Toscana per il 2013, 2014 e 2015 utilizzando il modello regressivo (giorno dell’anno previsto dal modello Stevenson e media della temperatura da Worldclim da Marzo ad Agosto per ogni anno; vedi paragrafo Data Spatialization per la funzi-one). L’intervallo del giorno previsto è indicato con i colori in legenda.

ciano 3-17.indb 20 24/11/17 11.49

21

Ital

ian

Jour

nal o

f Agr

omet

eoro

logy

- 3/

2017

Riv

ista

Ita

liana

di A

grom

eteo

rolo

gia

- 3/2

017

fields in Tuscany (data not shown). This implies that the accuracy of degree-day models could be significantly enhanced by biofix that corresponds to predictable biological events rather than those using fixed standard dates, such as 1 January. The inclusion of a set biofix (the first adult caught in field and/or the planting date) is recommended to bring the model output near the real observed day. The good performance of Stevenson model confirmed that WCR univoltine life history is closely synchronized with the phenology of maize, larval stages and maize root development being connected in field conditions (Bergman and Turpin, 1986; Naranjo, 1991), because of common dependence on environmental temperatures. With Stevenson model, factors affecting the start of emergence may be further reduced by using the planting date of maize as biofix, as well as the factors that affect heat unit accumulation by the immature stages, such as soil type, soil moisture, and the vertical distribution of eggs in the soil. Ellsbury and Lee (2004) suggested that wetness and temperature may both influence overwintering survival of WCR, the lack of winter precipitation leading to a high mortality rate of eggs close to the soil surface.In our study, we took advantage of phenological models implemented for other locations, where an accurate assessment of the developmental threshold and of the date after which development of WCR becomes temperature-dependent have already been assessed. More efficient predictions of regional differences in WCR development based on air temperatures would enhance integrated pest management efficacy (e.g., Davis et al., 1996). Further development could be represented by implementing the models with WCR distribution in different crop growing stages. Indeed, a correlation between WCR adult distribution and the density distribution of flowering maize was observed by Toepfer et al. (2007). However, increasing the complexity of degree-day based phenological models, by including a larger number of factors that can affect pest development, does not ensure greater accuracy (Damos and Savopoulou-Soultani, 2010). The simplicity and user-friendliness of the proposed method (and of its ensuing model), and the requirements of only daily maximum and minimum temperatures at the nearest weather station (and the date of planting), make this approach especially useful with a view to the integrated management of WCR populations and for alerting farmers through technical bulletins.In conclusion, Stevenson model appeared to be flexible enough to provide in season adjustment for

for designing local and regional pest management policies for maize crop in Tuscany, and for the selection of monitoring sites and tools. Among the others, the monitoring costs vary with the level of infestation and the size of the area, and the characteristics of the monitoring tools. Therefore, maps of potential pest distributions are urgently needed for designing alternative options. Although the scouting for adults may not be eliminated, the use of models will further improve the monitoring efficiency by allowing growers to focus on key periods, such us peak WCR emergence, when populations should be at their maximum in the field. De Luigi et al. (2011) conducted a geostatistical analysis that was proved to be effective in mapping the spread of WCR in Veneto Region (northern Italy), and concluded that the large-scale pattern of WCR dispersion can be accurately described by the spatial approach, thus optimizing the monitoring and subsequent control of this pest. The earliest and latest dates of the onset of adult emergence were separated by 20 days, across the three areas and the three years. In the same year, the differences among the three areas were 5 days in 2013, 20 days in 2014, and 12 days in 2015. The highest values of 2014 were probably dependent on the seasonal weather trend, starting from January. Summer 2014 was unusually cool and moist, while summer 2015 was relatively hot. The previous winter was exceptionally mild, both in 2014 and 2015. This trend could explain the early set of adult emergence observed in 2014. The earliest peak of adult flight occurred on 12 July, and the latest date of peak emergence was 8 August 2014. The earliest and latest dates at which the peak of adult emergence occurred were separated by 27 days. WCR emergence can peak 18 days after the onset and complete emergence can follow 53 days after the starting. The period for total beetle emergence observed in this study was very similar to what reported by other authors. Fisher (1984) counted 50 days for the complete emergence, starting from the first catch, whereas Nowatzki et al. (2002) observed 53 days for the complete emergence and Ruppel et al. (1978) 58 days. Park and Tollefson (2005) demonstrated that adult WCR distributions at peak population densities provided the best spatial prediction for the potential of larval damage to maize the following year, suggesting that the sampling of WCR adults can be considered an economic and effective predictor of the infestations.The model developed by Davis et al. (1996) with biofix at the beginning of the year was not able to predict emergence in any of the monitored

ciano 3-17.indb 21 24/11/17 11.49

22

Ital

ian

Jour

nal o

f Agr

omet

eoro

logy

- 3/

2017

Riv

ista

Ita

liana

di A

grom

eteo

rolo

gia

- 3/2

017

Chandler L.D., Faust R.M., 1998. Overview of areawide management of insects. Journal of Agricultural Entomolohy, 15: 319-325.

Ciosi M., Miller N.J., Kim K.S., Giordano R., Estoup A., Guillemaud T., 2008. Invasion of Europe by the western corn rootworm, Diabrotica virgifera virgifera: multiple transatlantic introductions with various reductions of genetic diversity. Molecular Ecology, 17: 3614-3627.

Clark T.L., Hibbard B.E., 2004. Comparison of nonmaize hosts to support western corn rootworm (Coleoptera: Chrysomelidae) larval biology. Environmental Entomology, 33: 681-689.

Damos P.T., Savopoulou-Soultani M., 2010. Development and statistical evaluation of models in forecasting moth phenology of major lepidopterous peach pest complex for Integrated Pest Management programs. Crop Protection, 29: 1190-1199.

Davis P.M., Brenes M.L., Allee L.L., 1996. Tem-perature dependent models to predict regional differences in corn rootworm (Coleoptera: Chrysomelidae) phenology. Environmental Entomology, 25: 767-775.

De Luigi V., Furlan L., Palmieri S., Vettorazzo M., Zanini G., Edwards C.R., Burgio G., 2011. Results of WCR monitoring plans and evaluation of an eradication programme using GIS and Indicator Kriging. Journal of Applied Entomology, 135: 38-46.Dixon A.F.G., Honek A., Keil P., Kotela M.A.A.A., Sizling, A.L., Jarosik V., 2009. Relationship between the minimum and maximum temperature thresholds for development in insects. Functional Ecology, 23: 257-264.

Ellsbury M.M., Lee R.E.J., 2004. Supercooling and cold-hardiness in eggs of western and northern corn rootworms. Entomologia Experimentalis et Applicata, 111: 159-163.

Estes R. E., Tinsley N. A., Gray M. E., 2016. Evaluation of soil-applied insecticides with Bt maize for managing corn rootworm larval injury. Journal of Applied Entomology, 140: 19-27.

Fisher J.R., 1984. Comparison of emergence of Diabrotica virgifera virgifera (Coleoptera: Chry- somelidae) from cut and uncut corn plants in artificial and natural infestations. Journal of the Kansas Entomological Society, 57: 405-408.

Fisher J.R., Sutter G.R., Branson T.F., 1990. Influence of corn planting date on the life stage development and phenology of Diabrotica virgifera virgifera. Entomologia Experimentalis et Applicata, 54: 219-224.

early and late planting, late emergence of maize seedlings, taking into account the effects of drought and heat stress on WCR food sources. One direct use of this model at the regional level is for identifying the start, maximum emergence point and end of the flight period to optimize the design and planning of population sampling, monitoring and/or control measures, with substantial savings in material and human resources. Although the distance between maize fields and the phenological status of maize may influence inter-field pest movements only to a certain extent, the combination of these indicators with population density of WCR may improve the predictive power of broad-scale precision modelling of landscape level pest dynamics. A continuous monitoring of the interactions between population dynamics and landscape variables associated with varying WCR density patches appears, once again, important for developing spatially-explicit DSS for AWPM, thus supporting advisors and farmers with data that help decide how, where and when managing this pest in Tuscany.

ReFeRencesAllen C., 1976. A modified sine wave method

for calculating degree days. Environmental Entomology, 5: 388-396.

Baker C.R.B., 1991. The validation and use of a life-cycle simulation model for risk assessment of insect pests. EPPO Bulletin, 21: 615-622.

Beckler A.A., French B.W., Chandler L.D., 2004. Characterization of western corn rootworm (Coleoptera: Chrysomelidae) population dynamics in relation to landscape attributes. Agricultural and Forest Entomology, 6: 129-139.Bergman M.K., Turpin, F.T., 1986. Phenology of field populations of corn rootworms (Coleoptera: Chrysomelidae) relative to calendar date and heat units. Environmental Entomology, 15: 109-112.

Boetel M.A., Fuller B.W., Chandler L.D., Hovland D.G., Evenson P.D., 1998. Fecundity and egg viability of northern and western corn rootworm (Coleoptera: Crysomelidae) adults surviving labeled and reduced soil insecticide applications. Journal of Economic Entomology, 91: 274-279.

Boriani M., Agosti M. Kiss J., Edwards C.R., 2006. Sustainable management of the western corn rootworm (Diabrotica virgifera virgifera LeConte, Coleoptera: Chrysomelidae) in infested areas: Italian, Hungarian and USA experiences. OEPP/EPPO Bulletin, 36: 531-537 .

ciano 3-17.indb 22 24/11/17 11.49

23

Ital

ian

Jour

nal o

f Agr

omet

eoro

logy

- 3/

2017

Riv

ista

Ita

liana

di A

grom

eteo

rolo

gia

- 3/2

017

and spatial dynamics of Bactrocera oleae (Rossi) infestations using decade-long agro-meteorological time series. International Journal of Biometeorology, 60: 1681-1694.

Meinke L.J.,Sappington T.W.,Onstad D.W., Guillemaud T.,Miller N.J.,Komáromi J.,Levay N.,Furlan L.,Kiss J.,Toth F., 2009. Western corn rootworm (Diabrotica virgifera virgifera LeConte) population dynamics. Agricultural and Forest Entomology, 11: 29-46.

Miller N., Estoup A., Toepfer S., Bourguet D., Lapchin L., Derridj S., Kim K.S., Reynaud P., Furlan L., Guillemaud T., 2005. Multiple transatlantic introductions of the western corn rootworm. Science, 310: 992.

Moeser J., Guillemaud T., 2009. International cooperation on western corn rootworm ecology research: state-of-the-art and future research. Agricultural and Forest Entomology, 11: 3-7.

Naranjo S.E., 1991. Movement of corn rootworm beetles, Diabrotica spp. (Coleoptera: Chry-somelidae), at cornfield boundaries in relation to sex, reproductive status, and crop phenology. Environmental Entomology, 20: 230-240.

Niccolai M., Marchi S., 2005. Il clima della Toscana. RaFT 2005:Rapporto sullo Stato delle Foreste in Toscana–Sherwood 124 (7/06), suppl. n. 2, pp 16-21.

Nowatzki T.M., Tollefson J.J., Calvin A.D., 2002. Development and validation of models for predicting the seasonal emergence of Corn Rootworm (Coleoptera: Chrysomelidae) beetles in Iowa. Environmental Entomology, 31: 864-873.

Oyediran I.O., Hibbard B.E., Clark T.L., 2004. Prairie grasses as hosts of the western corn rootworm (Coleoptera: Chrysomelidae). Environmental Entomology, 33: 1497-504.

Park Y.L., Tollefson J.J., 2006. Development and economic evaluation of spatial sampling plans for corn rootworm Diabrotica spp. (Col., Chry- somelidae) adults. Journal of Applied Ento-mology, 130: 337-342.

Petacchi R., Marchi S., Federici S., Ragaglini G., 2015. Large-scale simulation of temperature-dependent phenology in wintering populations of Bactrocera oleae (Rossi). Journal of Applied Entomology, 139: 496-509.

Régnière J., 1996. A generalized approach to land- scape-wide seasonal forecasting with temperature-driven simulation models. Environmental Entomology, 25: 869-81.

Riedell W.E., Reese R.N., 1999. Maize morphology and shoot CO2 assimilation after root damage by

Furlan L., Canzi S., Di Bernado A., Edwards C.R., 2006. The ineffectiveness of insecticide seed coatings and planting-time soil insecticides as Diabrotica virgifera virgifera LeConte population suppressors. Journal of Applied Entomology, 130: 485-490.

Godfrey L.D., Meinke L.J., Wright R.J., 1993. Vegetative and reproductive biomass accu-mulation in field corn - response to root injury by western corn-rootworm (Coleoptera, Chry-somelidae). Journal of Economic Entomology, 86: 1557-1573.

Gassmann A.J., Petzold-Maxwell J.L., Keweshan R.S., Dunbar M.W., 2011. Field evolved resistance to Bt maize by western corn rootworm. PLoS One, 6: e22629.

Gray M.E., Steffey K.L., Oloumi-Sadeghi H., 1993. Participatory on-farm research in Illinois cornfields: An evaluation of established soil insecticide rates and prevalence of corn rootworm (Coleoptera: Chrysomelidae). Journal of Economic Entomology, 86: 1473-1482.

Gray M.E., Felsot A.S., Steffey K.L., Levine E., 1992. Planting time application of soil insecticides and western corn rootworm (Coleoptera: Chrysomelidae) emergence: implications for longterm management programs. Journal of Economic Entomology, 85: 544-553.

Hijmans R.J., Cameron S.E., Parra J.L., Jones P., Jarvis A., 2005. Very high resolution interpolated climate surfaces for global land areas. International Journal of Climatology, 25: 1965-1978.

Kehlenbeck H., Krügener S., 2014. Costs and benefits of plant health measures against Diabrotica: experiences and estimations for Germany. Journal of Applied Entomology, 38: 222-233.

Kumar S., Graham J., West A.M., Evangelista P.H., 2014. Using district-level occurrences in MaxEnt for predicting the invasion potential of an exotic insect pest in India. Computers and Electronics in Agriculture, 103: 55-62.

Levine E., Spencer J.L., Isard S.A., Onstad D.W., Gray M.E., 2002. Adaptation of the western corn rootworm to crop rotation: Evolution of a new strain in response to a management practice. American Entomologist, 48: 94-107.

Liebhold A.M., Rossi R.E., Kemp W.P., 1993. Geostatistics and geographic information systems in applied insect ecology. Annual Review of Entomology, 38: 303-327.

Marchi S., Guidotti D., Ricciolini M., Petacchi R., 2016. Towards understanding temporal

ciano 3-17.indb 23 24/11/17 11.49

24

Ital

ian

Jour

nal o

f Agr

omet

eoro

logy

- 3/

2017

Riv

ista

Ita

liana

di A

grom

eteo

rolo

gia

- 3/2

017

N.C., 1991. Effect of insecticide treatments on survival, development, fecundity, and sex ratio in controlled infestations of western corn rootworms (Coleoptera: Chrysomelidae). Journal of Economic Entomology, 84: 1905-1912.

Szalai M., Komaromi J.P., Bazok R., Barcic J.I., Kiss J., Toepfer S., 2011. Generational growth rate estimates of Diabrotica virgifera virgifera populations (Coleoptera: Chrysomelidae). Journal of Pest Science, 84: 133-142.

Toepfer S., Ellsbury M.M., Eschen R., Kuhlmann U., 2007. Spatial clustering of Diabrotica virgifera virgifera and Agriotes ustulatus in small-scale maize fields without topographic relief drift. Entomologia Experimentalis et Applicata, 124: 61-75.

Tollefson J.J., 1998. Rootworm areawide management program in Iowa. Journal of Agricultural Entomology, 15: 351-357.

Urías-López M.A., Meinke L.J., 2001. Influence of western corn rootworm (Coleoptera: Chrysomelidae) larval injury on yield of different types of maize. Journal of Economic Entomology, 94: 106-111.

Witmer J.E., Hough-Goldstein J.A., Pesek J.D., 2003. Ground dwelling and foliar arthropods in four cropping systems. Environmental Entomology, 32: 366-376.

western corn rootworm larvae. Crop Science, 39: 1332-1340.

Rondon S.I., Gray M.E., 2003. Captures of western corn rootworm (Coleoptera: Chrysomelidae) adults with Pherocon AM and vial traps in four crops in east central Illinois. Journal of Economic Entomology, 96: 737-747 .

Ruppel R.F., Russell H.L., Jennings S.J., 1978. Indices for projecting emergence of corn rootworm adults in Michigan. Journal of Economic Entomology, 71: 947-949 .

Schroeder J.B., Ratcliffe S.T., Gray M.E., 2005. Effect of four cropping systems on variant western corn rootworm (Coleoptera: Chrysomelidae) adult and egg densities and subsequent larval injury in rotated maize. Journal of Economic Entomology, 98: 1587-1593.

Spencer J.L., Hibbard B.E., Moeser J., Onstad D.W., 2009. Behaviour and ecology of the western corn rootworm (Diabrotica virgifera virgifera LeConte) (Coleoptera: Chrysomelidae). Agricultural and Forest Entomology, 11: 9-27.

Stevenson D.E., Michels G.J., Bible J.B., Jackman J.A., Harris M.K., 2008. Physiological time model for predicting adult emergence of western corn rootworm (Coleoptera: Chrysomelidae) in the Texas High Plains. Journal of Economic Entomology, 101: 1584-1593.

Sutter G.R., Branson T.F., Fisher J.R., Elliott

ciano 3-17.indb 24 24/11/17 11.49