Biol Invasions (2014) 16:1851–1863 DOI 10.1007/s10530-013-0631-8

ORIGINAL PAPER

Light brown apple ( postvittana) (: ) colonization of

D. M. Suckling • L. D. Stringer • D. B. Baird • R. C. Butler • T. E. S. Sullivan • D. R. Lance • G. S. Simmons

Received: 29 March 2012 / Accepted: 21 December 2013 / Published online: 5 January 2014 Ó Springer Science+Business Media Dordrecht 2014

Abstract In response to the confirmed detection of for each trap based on the stated servicing period the , Epiphyas postvittana,in (*30,000), before analyzing for trends. Residual data California, approximately 53,000 pheromone-baited error rate was estimated as 1.5 %. and Jackson traps were deployed and more than 246,000 Santa Cruz counties had relatively high trap catches males were caught (February 2007–February 2010). immediately upon trap set, and remained the leading Approximately 46,000 manually entered catch records population centers, while most other counties showed were corrected for errors and converted into catch per a more general trend of a slow build-up in catch over trap per day. As empty trap data (zeros) were not time (12 counties). An exponential increase in trap recorded, we added zeros between first and last catch catch was observed in four counties with sufficient data. The pattern of spread indicated natural, as well as anthropogenic-assisted spread rates, with populations appearing well ahead of the invasion front. This jump dispersal is probably due to movement of host plants, unsurprising since eggs of this polyphagous moth are Electronic supplementary material The online version of readily laid on foliage. There was evidence of this article (doi:10.1007/s10530-013-0631-8) contains supple- seasonality in spread, probably linked to the phenol- mentary material, which is available to authorized users. ogy of the . There was a positive relationship D. M. Suckling (&) Á L. D. Stringer Á between catch and known host tree preference, R. C. Butler Á T. E. S. Sullivan suggesting that trap placement in preferred hosts The Institute for Plant and Food Research could add sensitivity to future surveys. Recommen- Limited, PB 4704, Christchurch 8140, New Zealand dations include the improved provision of data acqui- e-mail: [email protected] sition by telecommunications, standardization of data D. B. Baird input, more archiving, and frequent analysis of trap 8 Mariposa Crescent, Aidanfield, Christchurch 8025, catches. The rapid rate of population growth demon- New Zealand strated in two counties and spread across many others D. R. Lance supports the hypothesis of the recent arrival of E. Otis Laboratory, USDA-APHIS-PPQ-CPHST, postvittana in California. Buzzards Bay, MA 02542, USA Keywords Invasion Á Incursion Á Pheromone G. S. Simmons USDA-APHIS-PPQ-CPHST, Moss Landing, trap Á Population spread Á Delimitation Á CA 95039-9672, USA Leafroller 123 1852 D. M. Suckling et al.

Introduction (Suckling and Brockerhoff 2010). In addition to international restrictions, a U.S. federal order for The number of border incursions and government California established a list of regulated articles that eradication programs are increasing worldwide. could not be moved interstate from defined quarantine Together these highlight the significant costs of areas without official determination that such move- globalization (Tobin et al. 2013). Active trapping ment presented no risk of spreading LBAM. As a pest, systems, such as pheromone traps, have revolutionized LBAM is best known from fruit crops, and to a lesser pest management and border biosecurity by enabling extent from forestry, vegetable, and flower crops efficient detection and sampling of insect populations. (Wearing et al. 1991). It has been found on foliage of The analysis of large datasets produced from extensive many species in the large domestic export nursery population samplings have provided much valuable industry in California and occurs on annual weeds and information, leading to improved pest management or shelter tree species. The probable host list includes eradication outcomes. Pheromone traps have been the *500 species (Brockerhoff et al. 2011), and this is principal means for delimitation, population estima- likely to have led to this invasive insect becoming one tion and spread of the invasive gypsy moth, Lymantria of the most abundant micro-lepidopterans in New dispar (Sharov et al. 2002). Trap data have provided Zealand and the United Kingdom. In the southeastern valuable information to managers, enabling accurate United Kingdom, LBAM showed limited geographic targeting of aerial applications of sex pheromones for expansion for several decades, until the movement of , or of targeted insecticides to slow nursery stock became common practice in the 1980s, the spread of this species in North America (Sharov coinciding with its rapid spread (Porter 2001). Modes et al. 2002). Recently, Bigsby et al. (2011) highlighted of dispersal include ballooning of larvae as well as the importance of determining all modes of dispersal, female flight, and passive transport of all life stages by showing that quarantine regulations at the industry with host material (Brockerhoff et al. 2011). The rate level were successful at reducing gypsy moth spread. of localized or natural spread appears to be limited, However, it was unsuccessful at reducing the spread because of the relatively limited natural dispersal from non-commercial activities, such as firewood ability of females, at well under 1 km per generation collection by the public, a major factor promoting (Suckling et al. 1994). spread of gypsy moth. For the area-wide eradication Market access issues related to LBAM are of campaign against the pink bollworm, Pectinophora considerable importance in and New Zea- gossypiella, population monitoring with pheromone land (Suckling and Brockerhoff 2010). The zero traps has been essential for decision support and tolerance of live larvae in exports significantly raises timing of control measures when evaluating program the requirement for control in the field (Varela et al. progress both in real time and between seasons 2010). Market access issues also prompted the estab- (Henneberry 2007). Similarly, the Canadian codling lishment of a response program when LBAM was moth, Cydia pomonella, population suppression pro- detected in California in 2006 (Brown et al. 2010). gram has used pheromone traps to provide evidence Pheromone trapping commenced in Berkeley, CA, in for the successful population reduction in areas March 2007 with approximately two traps per km2 affected (Bloem et al. 2007). (five traps per square mile) in many counties across the Epiphyas postvittana, the light brown apple moth state (Brown et al. 2010). The detection and delimi- (LBAM, Lepidoptera: Tortricidae), native to Australia tation program eventually involved *53,000 traps, (Geier and Briese 1981; Fowler et al. 2009), was first making it one of the largest moth-trapping programs identified from a light trap in California in 2006 ever conducted. (Brown et al. 2010). It was determined as a high-risk The extensive trapping network to delimit the pest by the California Department of Agriculture extent of the LBAM infestation was critical to (Johnson et al. 2007), because it presented a market informing response managers of the pest’s distribution access barrier for export of fresh produce to many and to allow for the spatial positioning of quarantine countries that were free of this pest. It triggered a measures to mitigate its spread. The goals of this study major incursion response (initially $74M USD) by the were to estimate the observed combined rate of natural U.S. government at both federal and state levels and human-assisted spread of LBAM during its 123 Light brown apple moth (Epiphyas postvittana) 1853 colonization of California, and to derive lessons San Francisco, San Mateo, Santa Clara, Santa Cruz, leading to improvements to the LBAM regulatory and Solano counties. Numbers of traps deployed and quarantine efforts in California and elsewhere. An increased monthly during 2007 (April: 4,383, May: additional aim was to inform the incursion response 9,619, June: 14,526, July: 18,355, August: 24,164), managers, to improve responses to other invasive pest after which trap numbers increased more slowly, to organisms requiring large-scale surveillance, such as reach 53,084 in 39 counties by December 2008. Lobesia botrana (Lepidoptera: Tortricidae), a recent Starting from 19 July 2009 in San Mateo, Santa Cruz, arrival to California (Varela et al. 2010). Monterey and Contra Costa counties, most or all traps from detection grids were removed from areas that were known to be infested. This enabled traps to be Methods better used at the perceived invasion front, but prevented further analysis here after this date. Data Jackson traps, made of plastic-coated cardboard with a recorded for each trap included the latitude and sticky base (12.7 by 9.5 cm), were baited with longitude, street address, tree or plant in which the halobutyl septa (West Pharmaceutical, Lionville trap was placed, some trap retrieval dates, trapping PA), loaded with a two-component blend of phero- density, and counts of LBAM. Empty traps were not mone [96 % (E)-11-tetradecenyl acetate and 4 % recorded. (E,E)-9,11-tetradecadien-1-yl acetate (Bellas et al. 1983)] to trap male E. postvittana. Lure formulation was performed by a USDA-APHIS laboratory (Buz- Analysis of trap records zards Bay, MA, USA) until late 2007, after which a commercial source was used (Suterra LLC, Bend, OR, The 46,971 positive trap records (at least one LBAM) USA). To keep the traps functioning properly, the from April 2007 to April 2010 were supplemented pheromone-baited septa were replaced every 6 weeks. with zeros, generated by back-filling the data between Sticky inserts were replaced when either LBAM were first and last catch for each trap. Because traps that trapped on a base, which was then collected for never trapped LBAM were not recorded in the identification, or every month, as traps became dust or datasheet, and data on the timing and of placement debris-covered. During the incursion response, one- of traps were missing for many of the traps, we square-mile trapping grids (*2.6 km2) were assigned decided to concentrate our analysis on traps that we detection or delimitation status. Each were confident were present in a known grid, rather 1.6 km 9 1.6 km detection grid was assigned five than introducing more potential sources of error. We pheromone traps that were checked every 14 days for have certainly missed a large number of zeros by . Traps were rotated every 6 weeks, so that leaving out the traps that do not exist in the database. every 30 weeks the same location was surveyed twice. Our approach gave a total data set of 76,960 records If LBAM were detected, delimitation trapping was (traps with and without catch, far below what would be implemented whereby the grid where LBAM was expected from 53,000 traps checked fortnightly). The detected contained 100 traps and the surrounding eight increasing deployment of traps over the first few grids assigned 25 traps each. Traps were checked months is compensated for by calculation of catch per weekly for delimitation or biweekly for detection trapping effort (catch per trap per day). The data were grids; results were converted to catch per trap per day. subjected to quality assurance, with incomplete In March 2007, pheromone traps were loaded with records discarded and obvious (e.g. spelling) errors 1 mg pheromone-baited septa. In April 2007, phero- fixed. Records were checked so that the dates and mone lure loading was increased to 3 mg per trap to information recorded were consistent across time and increase trap sensitivity (Suckling and Karg 1996), matched the patterns of collection over varying which is desirable for detection and delimitation. Our intervals (Supplementary Fig. S1). The length of the analysis started from April 2007, when traps were capture period (trapping days) was used in the model deployed for initial detection trapping at a density of as an ‘‘offset’’ to scale the catches to a per day basis. five traps per mile2 (*two traps per km2) in Alameda, The ‘‘offset’’ is a technical term in generalized linear Contra Costa, , Marin, Monterey, Napa, models for a covariate that has a fixed coefficient of 1. 123 1854 D. M. Suckling et al.

This allowed us to adjust means over different periods moth just arrived in Santa Cruz, we investigated possible to have a constant rate per day, whilst allowing for the reasons for this by looking at the rate of spread and fact that data from longer periods have a different potential population growth. variance from those from shorter periods. For a subset of the data (March 2007–June 2009, To investigate population spread, trap locations with 92,714 trapped from 12,935 records), we were converted from latitude and longitude to Eastings had information on the plants in which the traps were and Northings for spatial analysis of the spread in placed. To determine the effect of plant on trap catch, meters. Eastings and Northings were calculated using we compared catches between traps placed in different the Universal Transverse Mercator coordinate system plants. Plant names (483 host descriptors) recorded as for the U.S. west coast. Northings are distances in supporting traps were categorized to family and order. meters from the equator, and Eastings are distances in Plants supporting traps were then categorized by rank meters from the -117° meridian. Change in catch over of host preference on a three-step scale of suitability time was examined separately for Eastings and (very common host, common host and occasional Northings to investigate spread in a single direction host) from Brockerhoff et al. (2011). The relationship and to reduce background noise. Smoothed curves between trap catch and host preference was analyzed illustrated in the figures used a cubic smoothing spline with a Poisson generalized linear model (McCullagh fitted in a generalized linear model with Poisson and Nelder 1989), which included host status as a errors. The Poisson errors allow for larger counts three-level factor and log(trapping days) as an offset, having higher variance than low counts. The smooth- to give an assessment of mean catch per trap per day. ing spline had 12 degrees of freedom, giving it plenty GoogleTM Earth Pro was chosen as the mapping of flexibility to follow the data, although this was system of choice because of the ability to share files reduced in some counties with shorter time series, to easily with collaborators. As the trapping area had been avoid over-fitting. In the south, where the geography divided into one-mile-square grids, these grids were was simpler, the median distance of captured males used to summarize the data, averaging the traps to per trap per day from the geocenter of the distribution *3,500 data points. Cumulative catch for each grid was (spatial-center of trapped moths from the first used to calculate catch per trap per square mile per year, 3 months) was calculated for each month. and known GPS coordinates for all traps within each Continuous nightly light trapping had been con- grid were averaged to approximate each grid center. ducted in Berkeley from the time of initial LBAM Custom icons were scaled, shaded, and elevated discovery in California (Brown et al. 2010). Trap proportionally to the log10 of the catch within each grid catches from the five pheromone traps in the same for each year to show differences in LBAM catch across square mile grid (Grid 4273) were compared with grids over time. XML code was inserted between data catches from the light trap to compare the ability of the columns in the MicrosoftÒ Excel spreadsheet of LBAM two trapping methods to determine LBAM population catch and was then exported in. csv format. A text editor growth. Catches in traps covering a 6 9 10 km area that supports multi-line ‘search and replace’ and regular centered on the light trap at Berkeley were analyzed to expressions was used to strip unwanted quotes and look for area-wide effects over time. commas and to pad out the header and footer with style From 16 May to 31 October, 2005, California information and opening and closing tags. The file was deployed 860 LBAM pheromone traps across 18 counties saved with a. kml suffix and opened in GoogleTM Earth in conjunction with the Cooperative Pest Survey Pro- Pro, then was resaved as a. kmz file so that any images gram, a national effort sponsored by USDA-APHIS associated with the data set (e.g. custom icons or image (USDA 2011). The project used Pherocon IIC traps overlays) could be included in the zipped document, (Tre´ce´ Inc., Adair, OK, USA) with LBAM pheromone ready to be shared via email or other medium. lures (1 mg loading). No LBAM were detected in 2005. In 2007, a large population was detected by pheromone trapping with lures baited with 1 mg of pheromone in Results Santa Cruz, one of the areas monitored in 2005. As the population size was immediately larger than would be A total of 246,852 moths were caught in pheromone expected in the short timeframe between surveys had the traps during the period covered by this study (April 123 Light brown apple moth (Epiphyas postvittana) 1855

Fig. 1 Population increase (catch per trap per square mile per grids. Substantial populations were apparent in Santa Cruz (left) year) and spread of the light brown apple moth Epiphyas and San Francisco (right) in 2007 (top) and grew out from the postvittana in San Francisco and surrounding counties (looking centers over time. Peak catch of 1,288 moths per grid per year westward to the Pacific Ocean). The total moths caught per occurred at Golden Gate Park, right. Numbers are plotted per square mile in 2007, 2008, and 2009 are shown with county square mile at median trap positions for traps deployed at five boundaries in red, and trapped counties without catch in gray per square mile. Maps produced with GoogleTM Earth Pro

2007–April 2010) (Fig. 1). Catch was highest in the was observed within California, despite extensive urban centers of San Francisco (n = 80,667 moths), quarantine efforts targeting nurseries and other indus- Watsonville (n = 24,588 moths), Santa Cruz tries capable of supporting cryptic infestations of eggs, (n = 22,345 moths), and Berkeley (15,758 moths). larvae or pupae on foliage and fruits. Isolated popu- For most populations, trap catch increased and lations showed up at long distances from the core area, expanded after initial detection (Fig. 2). Traps in San which increased in size rapidly over the 3 years, filling Francisco caught 33 % of the total catch in the state in the empty spaces between populations as they from 2007 to 2010 and initially appeared to show a merged with each other (Fig. 1). more rapid rise in population size than seen elsewhere The next highest density populations were found in (Fig. 1). The other major and burgeoning population the neighboring counties of Alameda, Monterey, center was in Santa Cruz, which initially had a large Contra Costa, San Mateo, and Marin, all of which LBAM population in 2007 that exhibited a drop from showed a gradual build-up of LBAM over time the high initial catches, followed by gradual rise (Fig. 2). Counties further from San Francisco and similar to those in other counties, until the central traps Santa Cruz counties had fewer catches, although were removed in July 2009 (Fig. 2). A rapid spread catches at these locations also increased in frequency

123 1856 D. M. Suckling et al.

Fig. 2 Analysis of catch per trap per day of male light brown apple moth E. postvittana over time in 14 counties in California in Jackson traps baited with synthetic female moth sex pheromone. Data were corrected for zero catches after the first catch; X running along the base of the figures = one data point, indicating data density. Trap removal from E. postvittana population centers to invasion fronts occurred onwards from July 19, 2009 in San Mateo, Santa Cruz, Monterey and Contra Costa counties

and number over time. The exceptions to this pattern were three possible local extinctions, shown by a lack of catches for many months after initial low-level detections (Santa Barbara, San Luis Obispo, and Los Angeles). In the case of Santa Barbara, the application of mating disruption dispensers may have contributed to this extinction event. Unfortunately, the records available have not enabled retrospective analysis of the impact of such management efforts. Trap catch from 0.1 moths per trap per day to 1.0 moth per trap per day over 2 years in San Francisco and Alameda counties, where trapping in April 2007 indicated widespread occurrence already. If that increase is considered as occurring over 3.5 generations per year, as predicted from annual temperatures for San Francisco (Gutierrez et al. 2010), this would amount to a ten-fold increase in male moth catch in seven generations (or 1.4-fold increase per generation). The geocenter for San Fig. 3 Spread (km) and increase in catch per trap per day of Francisco (based on catch in the first 3 months) was light brown apple moth E. postvittana from San Francisco. Crosses indicate moth catch. The initial dip/drop in the peak at Golden Gate Park, a trapping hotspot throughout the *10 km represents the water (i.e. drop in the density of traps) program, where the high insect population was surrounding the San Francisco peninsula on three sides

123 Light brown apple moth (Epiphyas postvittana) 1857

Fig. 4 Rate of movement of the median front, from E. postvittana trap catch, with distance from San Francisco in 2007–2009 (derived from 50 % percentiles of monthly data used to generate the 6-monthly data in Fig. 3). Median distance of moth catch from Golden Gate Park, San Francisco (dashed lines), and linear regression weighted for moth catch (solid line) probably due to a high biodiversity of host plants and a regular irrigation regime, supporting successful and rapid population growth. The redeployment of traps from the epicenters (to the invasion fronts) is visible as a reduction in mean catch from July 2009 towards the end of the trapping period (Fig. 2). An increase in density of catches and spread of LBAM to the south was observed from the high populations in San Francisco into adjacent San Mateo County. The population increased gradually, as highlighted by the increased trap catch per day values over time in San Mateo County (Fig. 2). Fig. 5 Catch of light brown apple moth E. postvittana in the vicinity of Berkeley, California, over 3 years, expressed as catch Data were plotted as a smoothed curve of catch per per trap per day (C/T/D). Eastings and Northings are in meters. trap per day as a function of distance from the The map is centered on the area where the first detection E. geocenter at Golden Gate Park (Fig. 3). The GLM postvittana was made by the light trap a year earlier analysis of deviance tables for the San Francisco data indicated that the models of distance from the approximately 30 km from the San Francisco geocen- geocenter fitted to the counts were significant ter (Fig. 3), with population sizes growing rapidly but (P \ 0.001)(Supplementary Table 1). increasing slowly outwards. Examining catch over As San Francisco is surrounded by water, the San time and in various ways, two rates of spread were Mateo population is shown as the bottom of the dip at evident (Figs. 3, 4, 5). The first involved both short- approximately 10–20 km from the geocenter in Fig. 3, range (natural and locally human-assisted) diffusion- as it is the only area where traps could be placed. type processes exemplified by Berkeley (Fig. 5), while Spread from San Francisco into Marin, Contra Costa the longer distance jumps were probably human- and Alameda counties was also observed as the peak at assisted ([20 km, also seen as the long tail in Fig. 3).

123 1858 D. M. Suckling et al.

The effect of time on median distances from Golden Gate Park, San Francisco, to the expanding front of traps catching moths is displayed in Fig. 4. Fitting a linear regression weighted for the number of moths caught in each month gave a fitted model of median distance (km) = 1.73 ? 0.5 * (number of months from April 2007). This relationship accounts for 66 % of the variation of the data, and is significant at the 0.1 % level. This indicated that the moth population was dispersing at an average rate of half a kilometer per month. The variation around the fitted line was not random, with distance peaks occurring Fig. 6 Total light brown apple moth E. postvittana catch per around August–September. This is most probably year at Berkeley, California, in a residential light trap (bars), and in five pheromone traps (line) within the same square mile grid linked to the phenology of the moth, which appears to (Grid 4273) have overlapping generations at this time of the year in California (Buergi et al. 2011). Thus with the larger for the tree where the trap was placed population size, the likelihood of trapping a moth in (F2, 12,330 = 16.0; P \ 0.001 for an overall assess- the surveillance grid increases. ment of host differences). The greatest catch was in Moth catch in the pheromone traps in a 6 9 10 km traps placed in plants in the Rosales and Fabales area centered on the light trap at Berkeley (where traps (‘‘very common hosts’’) [0.668 (95 % confidence were deployed from April 2007) showed low catches interval 0.643, 0.693) males per trap per day], in the first year (Fig. 4), with 100 moths caught in the followed by common hosts [0.603 (0.586, 0.621)] five traps per square mile grid in 2007. The catch per and occasional hosts [0.576 (0.555, 0.597)]. trap per day mean (±SEM) in the grid rose from 0.075 (0.008) in 2007 to 0.136 (0.019) (or 1.8-fold by 2008) and 0.578 (0.081) or 4.24-fold in 2009, with some Discussion localized population peaks apparent. Catch in Berkeley showed a rapid rise each year The large-scale delimitation of LBAM using phero- after initial detection in both the backyard light trap mone traps deployed over 1,000 square miles of operated daily (Brown et al. 2010 and unpublished California showed rapid colonization of large areas in data) and in five pheromone traps within the same the first 3 years of trapping. San Francisco and Santa square mile neighborhood (Fig. 6). The number of Cruz counties were clearly identified as two popula- moths trapped in the light trap per year increased by tion epicenters, both of which already had well- 2.5-fold from 2006 to 2007, sixfold 2007 from to 2008, established populations at the time of discovery. There and 45-fold from 2008 to 2009. did not appear to be an increase in the population Light brown apple moth was present in compara- density from the initial high numbers at these tively high density over *16 square miles across locations. It is possible that traps were filled to Santa Cruz County when traps were placed out in capacity, saturated with LBAM, and were thus unable April 2007 (Fig. 7). The insect was not trapped soon to record increasing population size. However, 98.2 % enough in San Francisco for a direct comparison with of LBAM trapping records with the Jackson traps in Santa Cruz in April, although by the end of 2007 it had California involved fewer than 30 moths per trap, and colonized 47 of 48 square-mile grids over San only 1,394 records exceeded 30 moths per trap, with a Francisco and was heading south into San Mateo maximum catch per trap of 232 at a single reading. County. By the end of 2007, the population at Santa This suggests that trap saturation was not a general Cruz and Watsonville was contiguous over 50 km of problem. Apart from San Francisco and Santa Cruz, coastline, which was considerably more than in San trap catch over time indicated that populations Francisco (\20 km). expanded rapidly after initial detection at low numbers The analysis of mean trap catch by trap host plant (Fig. 1). The population growth was clearly rapid at showed that catch increased with the host preference most locations. Every stage of colonization was 123 Light brown apple moth (Epiphyas postvittana) 1859

Fig. 7 Comparison of catch per trap of E. postvittana at Santa Cruz, California, in traps baited with synthetic female sex pheromone in 2005 (yellow circles, CAPS program) in April when traps were first established, and in all of 2007 (red spheres) evident, from populations at high density, to detection Not all detected dispersal events appeared to result of incipient populations (visible as the top or bottom in self-sustaining populations, with low number of rows in Fig. 2, respectively). In 2007, 100 insects were catches in some traps ceasing after a number of weeks. caught in the 6 9 10 km trapping grid centered on the According to a genetic relatedness study by Tooman light trap in Berkeley with the first record, supporting et al. (2011), a separate incursion event appears to the notion that E. postvittana was widespread but have occurred in Los Angeles in 2007. Knowledge of possibly patchily distributed in the Berkeley area this is important, because the lack of catch for a around the light trap by that time. The rate of buildup sustained surveillance period (3 July 2007–20 July varied between years and locations, although in the 2009) may indicate possible lag times in population Berkeley grid 1.8-fold and 4.2-fold annual increases in growth (Crooks 2005), or highlight suboptimal delim- catch were seen for the same trapping effort in 2008 itation methodology for detecting small populations. It and 2009, respectively. is possible to distinguish various scenarios of coloni- It was possible to assess the spread rate of LBAM zation when there is considerable haplotype diversity from San Francisco. There was evidence of isolated self- (e.g. Tooman et al. 2011), but this is difficult when sustaining populations that were probably human- only small samples are taken from the native range assisted, with many of these populations occurring (Rubinoff et al. 2011). 15–20 km beyond population centers, a distance well At first glance, it appeared strange that an LBAM beyond the natural dispersal distance (Suckling et al. survey using pheromone traps in Santa Cruz in 2005 1994). The human dimension of insect dispersal has did not detect LBAM despite the insect being caught in been highlighted in the gypsy moth Lymantria dispar in large numbers immediately upon trap placement in North America, with jumps beyond normal dispersal April 2007. The following assumptions about trap- rates explained by public firewood collection rather than ping, heterogeneity on the landscape and population commercially based activities (Bigsby et al. 2011). growth may help to explain this. Firstly, 32-fold more

123 1860 D. M. Suckling et al. traps were present in 2007, reducing the amount of Berkeley in a single light trap and the five pheromone attraction-free space between pheromone traps. Sec- traps in the same square mile over three concurrent ondly, the population was potentially growing at an years showed good agreement between trapping estimated 1.4-fold increase in population size per methods. Both trap types detected LBAM from 2007 generation at 3.5 generations per year (Gutierrez et al. and revealed a similar dramatic upward trend in catch 2010) during the time between the two surveys, which thereafter (Fig. 6). The performance of the neighbor- was about four generations or *5.6-fold (December ing trap grids in the Berkeley area revealed that the 2005–April 2007). For comparison, at Berkeley the light trap was successful at detecting the moth, but annual population increases were 1.8- and 4.2-fold per LBAM had clearly been present in the area before year in 2007–2009. Taking into account an estimated 2006, based on the contiguous distribution seen in 5.6-fold lower population size at Santa Cruz and 2007 in pheromone trap grids. This situation enables 32-fold fewer traps, from the product of these the demonstration of rapid population growth at a components we can estimate a 178-fold lower likeli- single location, which is helpful because any estimate hood of catch in 2005 than in April 2007. Lower rates of the rate of spread of the insect is confounded by the of population increase, closer to 1.15-fold per gener- effects of the rate of population growth. Thus the ation, were measured in laboratory conditions, population median can in fact shift spatially inwards, depending on temperature and host plant (Danthan- if the population growth is more rapid in the center, or arayana et al. 1995), which would reduce this factor to outwards, if there is spread and more constant 146-fold, and using the lowest obtained field estimate population growth across its distribution. from Berkeley (1.8-fold growth in 2007–2008) would The availability of traps and lures greatly enhances reduce it to 66-fold higher surveillance efficiency in the rate of success of eradication programs (Tobin 2007 than in 2005. Spatial heterogeneity was also et al. 2013). Ideally, detection programs should be revealed by close range examination of traps at operated with the highest possible trap efficiency at Berkeley. Trapping grids did not all detect populations low insect density, but detection trapping is expensive. at the same time. The variation in trap servicing rate Improving our knowledge of the impact of trap could be a factor in some areas (Supplementary Fig. placement on trap sensitivity in relation to host status S1). Peaks are evident about the weekly returns at 7, or habitat has the potential to improve surveillance 14, 21 and 28 days, although some traps were checked efficiency without greatly increasing the costs, by less frequently. providing program managers with better guidance on There has been relatively little effort to compare where to place traps. In some species, sex pheromone light traps and pheromone traps, in part because of the trapping has been observed to attract more males when complexities in doing so, but it is worth considering presented with host plant odorants (Light et al. 1993; potential complicating factors. Delisle et al. (1998) Schmidt-Busser et al. 2009). Similarly, catch in traps found that for hemlock looper moths Lambdina placed in host trees may catch more than traps in non- fiscellaria fiscellaria, pheromone-baited traps outper- host trees (Suckling et al. 2007). This could be because formed light traps at low moth density, while the more insects are present in the host tree, or because of reverse occurred at high moth density. Light and a synergistic effect between the pheromone and host pheromone traps have been used in ecological studies odors. The trend for slightly higher LBAM catch in of heliothine moths (e.g. Hartstack et al. 1979; Baker preferred host trees suggests that trap placement in et al. 2011), but they have been used less frequently for host trees has the potential to enhance surveillance tortricids. A Japanese study on daily attraction rates to efficiency of E. postvittana pheromone traps. light traps found that only flight threshold tempera- Two modes of spread (natural and human-assisted) tures were limiting moth catch probability (Hirao et al. were confounded here. This is consistent with LBAM 2008). For LBAM, the flight threshold is 8–11 °C spread throughout New Zealand, the UK and else- (Danthanarayana 1976), and a quick search of histor- where (Suckling and Brockerhoff 2010). It is difficult ical meteorological data indicates that scotophase to predict what will happen in the future as LBAM temperatures in Berkeley between April and October increases its range in the USA, as outcomes vary with would not be limiting to LBAM flight (wunder- the modeling techniques employed (Gutierrez et al. ground.com). The comparison of catches in residential 2010; Lozier and Mills 2011). It may have economic 123 Light brown apple moth (Epiphyas postvittana) 1861 effects from limiting market access for exports, and/or four-component pheromone blend for LBAM (El- as it becomes more suited to its new range and or hosts, Sayed et al. 2011) is more attractive than females and it may become more of a pest (Simberloff 2009). the two-component synthetic blend used in California; A surprisingly high proportion of eradication it thus is likely to be more sensitive at detecting programs are successful (Tobin et al. early online), LBAM at low densities, particularly ahead of the and it is of interest here that the availability of lures invasion front, where populations are likely to be low. notably increases the prospects for success, more than More carefully planned and systematic testing of 20-fold. Eradication of insect pests over larger areas is ground-based eradication tactics, such as the applica- feasible, e.g. New World screw worm Cochliomyia tion of mating disruption (Suckling et al. 2012a, b) hominivorax was eradicated from the southern USA around incipient populations used to contain the through Mexico down to Panama (Klassen and Curtis population, would be useful to determine the value 2005; Wyss 2000). Eradication of E. postvittana might of this approach for future targets, including urban have been feasible soon after detection, but the rapid areas. Standard procedures between counties will spread reported here precluded this scenario. So, too, substantially improve the information that can be did the decision to avoid the use of aerial spray of provided for the program managers, and flexibility to pheromone, which meant that only ground-based use new tools will lead to better responses over time. control was possible. There have been a number of area-wide trapping Acknowledgments This work was funded by the US and delimitation programs reported, but only a few Department of Agriculture and the New Zealand Ministry for Science and Innovation (Better Border Biosecurity, www.b3nz. lepidopteran data sets have been considered in detail, org) and completed during a fellowship to DMS supported by such as gypsy moth (Bigsby et al. 2011; Sharov et al. the Organisation for Economic Cooperation and Development 2002), gum leaf skeletonizer Uraba lugens (Suckling (Cooperative Research Programme: Biological Resource Man- et al. 2005), and painted apple moth anartoides agement for Sustainable Agricultural Systems) at the European Biological Control Laboratory (USDA-Agricultural Research (Suckling et al. 2007; see Vreysen et al. 2007 for other Service), Montpellier, France. We thank Professor Jerry Powell insect examples). The value of analyzing such data for providing light trap data on light brown apple moth and the sets is potentially huge (Brockerhoff et al. 2010), with California Department of Food and Agriculture for their coop- the chance to improve future surveillance and eradi- eration. We thank Kathleen Harding, Laura Irons and Tim Horton for data mining and providing information on trapping cation programs greatly, reducing the long-term costs operations. and impacts of insect pests. An obvious recommendation to program managers would be to ensure that traps for LBAM detection are References placed in host trees, particularly those from the Orders Rosales and Fabales. Further, insect incursion Baker GH, Tann CR, Fitt GP (2011) A tale of two trapping response programs should ensure that there is frequent methods: Helicoverpa spp. (Lepidoptera, Noctuidae) in pheromone and light traps in Australian cotton production analysis of trap catch data, to provide real-time systems. Bull Entomol Res 101:9–23 decision support and to pick up on any errors. Bellas TE, Bartell RJ, Hill A (1983) Identification of two Systematic methods and protocols need to be used in components of the sex-pheromone of the moth, Epiphyas and across jurisdictions to ensure a unified and postvittana (Lepidoptera, Tortricidae). J Chem Ecol 9:503–512 consistent response, from data collection and record- Bigsby KM, Tobin PC, Sills EO (2011) Anthropogenic drivers ing, including collection of negative trap records, the of gypsy moth spread. Biol Invasions 13:2077–2090 operation dates of the traps, to consistent delimitation Bloem S, McCluskey A, Fugger R, Arthur S, Wood S, Carpenter and management protocols. Modern technologies J (2007) Suppression of the codling moth Cydia pomonella in British Columbia, Canada using an area-wide integrated (such as portable data acquisition devices) can offer approach with an SIT component. In: Vreysen MJB, significant advantages, such as reduced error rates for Robinson AS, Hendrichs J (eds) Area-wide control of data entry; use of these should be considered. insect pests: from research to field implementation. Furthermore, technologies are continually being Springer, Dordrecht, pp 591–601 Brockerhoff EB, Liebhold AM, Richardson B, Suckling DM improved, including new insect pheromone blends; (2010) Eradication of invasive forest insects: concept, thus program managers need to have the flexibility to methods, costs and benefits. N Z J For Sci 40(Sup- refine procedures as required. For example, a new pl.):S117–S135 123 1862 D. M. Suckling et al.

Brockerhoff EG, Suckling DM, Ecroyd CE, Wagstaff SJ, Raabe and potential impacts. UC Statewide Integrated Pest MC, Dowell RV, Wearing CH (2011) Worldwide host Management Program. http://www.aphis.usda.gov/plant_ plant use of the highly polyphagous, invasive Epiphyas health/plant_pest_info/lba_moth/downloads/LBAM_IPM_ postvittana (Lepidoptera: Tortricidae). J Econ Entomol UCDavis.pdf. Accessed 13 Jan 2012 104:1514–1524 Klassen W, Curtis CF (2005) History of the sterile insect tech- Brown JW, Epstein ME, Gilligan TM, Passoa SC, Powell JA nique. In: Dyck VA, Hendrichs J, Robinson AS (eds) (2010) Biology, identification and history of the light Sterile insect technique: principles and practice in area- brown apple moth Epiphyas postvittana (Walker) (Lepi- wide integrated pest management. Springer, Dordrecht, doptera: Tortricidae: ) in California: an example pp 3–36 of the importance of local faunal surveys to document the Light DM, Flath RA, Buttery RG, Zalom FG, Rice RE, Dickens establishment of exotic insects. Am Entomol 56:34–43 JC, Jang EB (1993) Host-plant green-leaf volatiles syner- Buergi LP, Roltsch WJ, Mills NJ (2011) Abundance, age gize the synthetic sex pheromones of the corn earworm and structure, and voltinism of light brown apple moth popu- codling moth (Lepidoptera). Chemoecology 4:145–152 lations in California. Environ Entomol 40:1370–1377 Lozier J, Mills N (2011) Predicting the potential invasive range Crooks JA (2005) Lag times and exotic species: the ecology and of light brown apple moth (Epiphyas postvittana) using management of biological invasions in slow-motion. biologically informed and correlative species distribution Ecoscience 12:316–329 models. Biol Invasions 13:2409–2421 Danthanarayana W (1976) Diel and lunar flight periodicities in McCullagh P, Nelder JA (1989) Generalized linear models, 2nd the light brown apple moth, Epiphyas postvittana (Walker) edn. Chapman & Hall, London (Tortricidae) and their possible adaptive significance. Aust Porter J (2001) Range expansion in the light brown apple moth J Zool 24:65–73 Epiphyas postvittana (Walk.). Atropos 14:1–4 Danthanarayana W, Gu H, Ashley S (1995) Population-growth Rubinoff D, Holland BS, San Jose M, Powell JA (2011) Geo- potential of Epiphyas postvittana, the lightbrown apple graphic proximity not a prerequisite for invasion: moth (Lepidoptera, Tortricidae) in relation to diet, tem- not the source of California invasion by light brown apple perature and climate. Aust J Zool 43:381–394 moth (Epiphyas postvittana). PLoS ONE 6:e16361 Delisle J, West RJ, Bowers WW (1998) The relative perfor- Schmidt-Busser D, von Arx M, Guerin PM (2009) Host plant mance of pheromone and light traps in monitoring the volatiles serve to increase the response of male European seasonal activity of both sexes of the eastern hemlock grape berry moths, Eupoecilia ambiguella, to their sex looper, Lambdina fiscellaria fiscellaria. Entomol Exp Appl pheromone. J Comp Physiol A Neuroethol Sens Neural 89:87–98 Behav Physiol 195:853–864 El-Sayed AM, Mitchell VJ, Manning LM, Suckling DM (2011) Sharov AA, Leonard D, Liebhold AM, Roberts EA, Dickerson New sex pheromone blend for the lightbrown apple moth, W (2002) ‘‘Slow the Spread’’: a national program to con- Epiphyas postvittana. J Chem Ecol 37:640–646 tain the gypsy moth. J For 100:30–35 Fowler G, Garrett L, Neeley A, Magarey R, Borchert D, Spears Simberloff D (2009) We can eliminate invasions or live with them. B. (2009) Economic analysis: risk to U.S. apple, grape, Successful management projects. Biol Invasions 11:149–157 orange and pear production from the light brown apple moth, Suckling DM, Brockerhoff EG (2010) Invasion biology, ecol- Epiphyas postvittana (Walker). http://www.aphis.usda.gov/ ogy, and management of the light brown apple moth plant_health/plant_pest_info/lba_moth/downloads/lbame (Tortricidae). Ann Rev Entomol 55:285–306 conomicanalysis.pdf. Accessed 13 Jan 2013 Suckling DM, Karg G (1996) Parameters affecting catch of Geier P, Briese D (1981) The light-brown apple moth, Epiphyas lightbrown apple moth in pheromone-treated orchards. postvittana (Walker): a native leafroller fostered by Proc 49th NZ Plant Prot Conf 49:319 European settlement. In: Kitching R, Jones R (eds) The Suckling DM, Brunner JF, Burnip GM, Walker JTS (1994) ecology of pests. CSIRO, Melbourne, pp 131–155 Dispersal of Epiphyas postvittana (Walker) and Plano- Gutierrez AP, Mills NJ, Ponti L (2010) Limits to the potential tortrix octo Dugdale (Lepidoptera: Tortricidae) at a Can- distribution of light brown apple moth in Arizona-Cali- terbury, New Zealand orchard. N Z J Crop Hort Sci fornia based on climate suitability and host plant avail- 22:225–234 ability. Biol Invasions 12:3319–3331 Suckling DM, Gibb AR, Dentener PR, Seldon DS, Clare GK, Hartstack AW, Witz JA, Buck DR (1979) Moth traps for the Jamieson L, Baird D, Kriticos DJ, El-Sayed AM (2005) tobacco budworm. J Econ Entomol 72:519–522 Uraba lugens (Lepidoptera: Nolidae) in New Zealand: Henneberry TJ (2007) Integrated systems for control of the pink pheromone trapping for delimitation and phenology. bollworm Pectinophora gossypiella in cotton. In: Vreysen J Econ Entomol 98:1187–1192 MJB, Robinson AS, Hendrichs J (eds) Area-wide control of Suckling DM, Barrington AM, Chhagan A, Stephens AEA, insect pests: from research to field implementation. Burnip GM, Charles JG, Wee SL (2007) Eradication of the Springer, Dordrecht, pp 567–579 Australian painted apple moth Teia anartoides in New Hirao T, Murakami M, Kashizaki A (2008) Effects of mobility Zealand: trapping, inherited sterility, and male competi- on daily attraction to light traps: comparison between tiveness. In: Vreysen MJB, Robinson AS, Hendrichs J (eds) lepidopteran and coleopteran communities. Insect Conserv Area-wide control of insect pests: from research to field Divers 1:32–39 implementation. Springer, Dordrecht, pp 603–615 Johnson MW, Pickel C, Strand LL, Varela LG, Wilen CA, Bolda Suckling DM, McLaren GF, Manning LM, Mitchell VJ, Attfield MP, Flint ML, Frankie Lam WK, Zalom FG (2007) Light B, Colhoun K, El-Sayed AM (2012a) Development of brown apple moth in California: quarantine, management, single-dispenser pheromone suppression of Epiphyas 123 Light brown apple moth (Epiphyas postvittana) 1863

postvittana, octo and obliquana Varela LG, Walker JTS, Lo PL, Rogers DJ (2010) New Zealand in New Zealand stone fruit orchards. Pest Manag Sci lessons may aid efforts to control light brown apple moth in 68:928–934 California. Calif Agric 64:6–12 Suckling DM, Sullivan TES, Stringer LD, Butler RC, Campbell Vreysen MJB, Robinson AS, Hendrichs J (2007) Area-wide DM, Twidle AM, Allen WJ, Mafra-Neto A, El-Sayed AM control of insect pests: from research to field implemen- (2012b) Communication disruption of light brown apple tation. Springer, Dordrecht moth (Epiphyas postvittana) using a four-component sex Wearing CH, Thomas WP, Dugdale JS, Danthanarayana W pheromone blend. Crop Prot 42:327–333 (1991) Tortricid pests of pome and stone fruits, Australian Tobin PC, Kean JM, Suckling DM, McCullough, DG, Herms, and New Zealand species. In: van der Geest LPS, Evenhuis DA, Stringer, LD (2013) Determinants of successful HH (eds) Tortricid pests: their biology, natural enemies and eradication programs. Biol Invasions. doi:10. control. World Crop Pests, vol 5. Elsevier, Amsterdam, 1007/s10530-013-0529-5 pp 453–472 Tooman LK, Rose CJ, Carraher C, Suckling DM, Paquette SR, Wyss JH (2000) Screw-worm eradication in the Americas- Ledezma LA, Gilligan TM, Epstein M, Barr NB, New- overview. In: Tan KH (ed) Area-wide control of fruit flies comb RD (2011) Patterns of mitochondrial haplotype and other insect pests. In: International conference on area- diversity in the invasive pest Epiphyas postvittana (Lepi- wide control of insect pests, and the 5th international doptera: Tortricidae). J Econ Entomol 104:920–932 symposium on fruit flies of economic importance 28 May– USDA (2011) Cooperative agricultural pest survey. http://www. 5 June 1998, Penang, Malaysia, 2000. Penerbit Universiti aphis.usda.gov/plant_health/plant_pest_info/pest_detection/ Sains downloads/survey/survey-2011/2011-NationalSurveyGuide lines.pdf. Accessed 13 Jan 2012

123