To Repeat: Can Polyphagous Invasive Tephritid Pest Populations Remain Undetected For Years Under Favorable Climatic and Host Conditions?

In recent decades, there has been a dramatic increase in detections of invasive pest tephritid fruit flies in California as well as other regions. Two camps have offered divergent explanations. One maintains that numerous introductions of maggots in infested fruit found incipient populations, which are detected by the existing trapping network and eradicated or, less frequently, not completely eradicated, a circumstance which is then followed up with another full set of eradication procedures (McInnis et al. 2017). Frequent detections, in this view, reflect increasing international human traffic and arrival of infested hosts, particularly in the Los Angeles megalopolis. Increased global movement of people and goods is also likely responsible for the increasing number of different fruit fly species being detected in California (from one species in 1960 to 17 species in 2017). The alternate view is that populations of several species, including the Mediterranean and oriental fruit flies, Ceratitis capitata (Wiedemann) and Bactrocera dorsalis (Hendel), respectively, have been established in the state for many years (Carey 1991, 1996, 2010; Papadopoulos et al. 2013), and detections result from occasional population build-ups in particular locations to the point where they are detected in traps. Once discovered, eradication procedures reduce, but do not eliminate, such populations to small, sub-detectable levels, where in time they may again grow to a detectable size. In the recent give-and-take, McInnis et al. (2017) challenged the validity of the “established population” argument by raising several obvious questions. If non-native pest fruit flies are established in California, why are they not trapped in rural areas, where the vast majority of commercial horticultural production is? Why have larvae never been detected in Californian produce by trading partners and foreign consumers? And why are detections in California so rare in spite of a continuously operating trapping network that includes more than 94,000 traps? Or, more precisely, what ecological or evolutionary factors act over long periods to maintain their populations at what Carey et al. (2017) now call “ultra-small” sizes as to be undetectable by trapping? Also, what are the specific conditions that trigger the sudden build-up of the populations that they claim remain undetectable for years? These fruit fly species are fecund, with relatively short generation times, and polyphagous, with hundreds of suitable host plants; they are highly mobile, and they have no apparent surfeit of natural enemies. In rebutting McInnis et al. (2017), Carey et al. (2017, hereafter referred to as the Carey Group) posited 10 rebuttal points that repeat previous arguments while failing to directly address the arguments of McInnis et al. (2017). We believe the Carey Group’s position rests largely on three assumptions: 1) there are long post-invasion lag times in population growth; 2) the existing trapping system for detecting non-indigenous tephritid species is ineffective; and 3) operational or regulatory international guidelines regarding eradication are weak compared to strictly science-based criteria. Accordingly, this essay focuses on these assumptions and comments briefly on the San Diego oriental fruit fly case highlighted by the Carey Group. Remarks on the remaining rebuttal points appear in Table 1. Carey et al.’s rebuttal points with corresponding responses. Successful invasions manifest a common process involving a series of sequential phases, namely transport, colonization, establishment, and spread (e.g., Blackburn et al. 2015). While the duration of these phases may vary, a temporal gap, or lag time, may occur between establishment and rapid population growth and spatial spreading (e.g., Mack et al. 2000). The Carey Group derides the expectation that non-native fruit flies in California should display rapid population growth, because they assume these species, despite their r-selected nature, are still in the lag phase of the invasion process and hence undetectable. However, lag phases, particularly for invasive insects, are generally not prolonged. The Carey Group noted that lag times may last decades or even centuries, but these extreme cases occur in long-lived, woody plants, such as certain oaks that do not produce acorns until 50+ years old (Kowarik 1995). Lag times vary substantially among taxa, being decades to centuries for woody plants (Kowarik 1995, Larkin 2012) and one to several decades for herbs (Pyšek and Prach 1993, Larkin 2012), birds (Aagaard and Lockwood 2014), and mammals (Crooks 2005), but only a few years for insects (Roques et al. 2016). Regarding insect invaders, Kenis et al. (2007) stated “In general, there is no obvious ‘lag phase’…and when a lag phase is observed, it is usually of short duration.” Examples of rapid population growth and spatial spreading are common for invading insects and include wasps (Bertolini et al. 2016), mosquitoes (Kaufman and Fonseca 2014), ladybird beetles (Lombaert et al. 2010), homopterans (Petit et al. 2008), and drosophilids (Roques et al. 2016), to mention a few. Rapid growth and spread also characterize invasive pest tephritid species. For example, B. dorsalis was discovered in Africa in 2003, but within three years, it was a major pest throughout the sub-Saharan region (Ekesi et al. 2006) and is now present in 41 African countries (De Meyer et al. 2014). In Australia, this same species invaded Queensland and rapidly spread to 20,000 km2 before an eradication program stopped the expansion and ultimately eliminated this very large infestation (Hancock et al. 2000). Rapid population growth and spread was also documented for medfly invading Guatemala and southern Mexico (Hendrichs et al. 1983). This same pest recently invaded the Dominican Republic, spreading in a few months to more than 2,000 km2 (Marte et al. 2017). Even more germane, and as noted by McInnis et al. (2017), the largely monophagous olive fly, Dacus oleae (Gmelin), was first detected in California in 1988 and within five years had spread to all olive-growing areas in the state. First discovery of medfly and oriental fruit fly in California occurred 42 (1975) and 57 (1960) years ago, respectively (Carey et al. 2017). To invoke the lag phase to explain the decades-long scarcity of these species in California is to propose delays in population growth of these highly fecund, mobile, and polyphagous insects that are, at minimum, several-fold longer than those typically observed for invasive arthropods and counter to documented cases in other parts of the world. Apparently, the Carey Group assumes that, because the invading tephritids are largely tropical in origin, they face an inhospitable climate in California, which constrains population growth. In other words, the long lag times are thought to represent time required for local adaptation. Regarding the medfly, it is interesting to note that Carey (1996) previously identified the southern coastal region of California, including the Los Angeles Basin and San Diego County, with a climate similar to Greece and Israel, where medfly is abundant, as prime habitat for the medfly and predicted that “all backyard hosts would eventually be infested with medfly throughout much of the year and commercial fruit-growing areas would be under constant attack.” It is not clear why this region of California, “with the most ideal conditions for medfly in the state” (Carey 1996), was previously considered eminently suitable for the medfly but now, apparently, is not. The high abundance of medfly in the Mediterranean region indicates that the species should similarly thrive in California. Regarding Bactrocera, the invasion history of the peach fruit fly, B. zonata (Saunders), to take one example, also casts doubt on the notion that California presents an “adaptive problem,” as this species has a tropical origin (Southeast Asia), yet has become established in Egypt, where it has become a serious agricultural pest (Ni et al. 2012). Similarly, B. dorsalis has rapidly adapted to a wide range of ecological zones (semi-arid to humid tropical) across the African continent (Goergen et al. 2011). Additionally, the strong likelihood (Barr et al. 2014) that non-native tephritids have been introduced to California on multiple occasions from multiple sources weakens the notion that “adaptive constraint” limits population growth. Repeated arrivals would serve to increase the genetic diversity of any established populations and thereby lessen the incidence and severity of genetic bottlenecks and increase the adaptive plasticity of the established populations (the “genetic rescue” effect, Carlson et al. 2014). Moreover, the theoretical expectation that adaptive potential is inversely related to population size may not be universally true (Hoffman et al. 2017), and indeed invading insects may evolve quickly to novel environments (e.g., Huey et al. 2000, Tanaka et al. 2015). Assuming that populations of non-indigenous tephritids have existed in California for decades implies that these flies have escaped detection by the ongoing extensive trapping program in the vast majority of times and places (94,000 detection traps serviced biweekly yields 2,444,000 trap checks per year). It also rests on the unlikely assumption that these populations are consistently small over long periods under favorable climatic and host conditions; otherwise, they would be detected within a few generations. In any case, the possibility of a small population not being detected at a given point in space and time has little bearing on whether broadly distributed populations of non-native tephritids persist in California. That question relates more specifically to whether populations can be large enough to be viable, display the characteristic demography and ecology of polyphagous invasive tephritids, and still escape the general detection trapping system in the vast majority of places and times. The maximum size that populations can achieve without being detected on a regular basis will depend on the sensitivity of the detection trapping system. Assessing that sensitivity requires an understanding of the relationship of trap capture to the numbers of flies in the field, and the latter can be difficult to measure, especially if populations are sparse. One approach is to release insects within a detection trapping network, such that the numbers of flies and their distances from traps are known, and then observe how many are caught. The proportions of flies captured can be used as estimates of the probability that a fly will be “detected” (caught) if present at a given distance from a trap (Lance and Gates 1994). Distance/capture functions based on mark-recapture data have been used to directly calculate probabilities of detecting C. capitata and Bactrocera populations of various sizes (Lance and Gates 1994, Shelly et al. 2010) and to develop Monte Carlo simulations of incipient tephritid populations within detection trapping grids (Lance 2014). Results of these exercises suggest that the detection grids in California, which include more than 94,000 traps that are regularly relocated according to host phenology, are quite sensitive. In high-risk areas (monitored at two traps per km2), the models predict that over half of the populations of C. capitata will be detected (at least one fly caught) by the time there are 250 males present, and more than 90% will be found at 2,000 males. For B. dorsalis, only about 30 males are required for a 50% detection probability, and the 90% level is reached at around 200 flies (Lance 2014). Because these estimates are based on probabilities of capture of individual flies, they apply across space and time. For example, assuming a population of C. capitata exists only in 10 “pockets” that are spread throughout “high-risk” areas of California—residential areas where the detections have historically occurred—we should expect to find at least one fly half of the time if the pockets average 25 males each, and 90% of the time if they average 200 males. To further illustrate, there have been many instances when two years have passed with no C. capitata detections in a given county. Assuming, conservatively, six to eight generations per year (Muñiz and Zalom 1997), these pockets of infestation would have to average two males or fewer per generation to avoid at least one system-wide detection 50% of the time, and 10 males or fewer per generation to escape 10% of the time. Given that populations in these size ranges are likely subject to Allee effects (Liebold and Tobin 2008) and will fluctuate both seasonally (Papadopoulos et al. 2001) and stochastically, we would argue that the long-term survival of a given population would be extremely unlikely under the scenario described here. A similar analysis with B. dorsalis, a species showing much higher attraction to traps, would show long-term survival of undetected populations to be even more unlikely when considering that B. dorsalis males seek out and ingest methyl eugenol, the bait used to trap these flies, in order to produce sex pheromone and to attract females for mating (Tan and Nishida 1996). Moreover, repeat captures of the same individuals in methyl eugenol traps (Tan and Jaal 1986) suggest that males may need to “re-fuel” on this pheromone precursor, which would increase the likelihood of detection. The Carey Group, however, dismiss the entire approach of using mark-recapture data for these analyses and state that “Detectability estimates based on results of release-recapture studies using tens of thousands of factory-reared laboratory strains of oriental fruit flies have little relevance to the detectability of wild flies living in their natural habitats.” Presumably, their argument would extend to similar studies with C. capitata (e.g., Lance and Gates 1994). In contrast, we argue that data from such tests are currently the best quantitative information available for estimating detection system sensitivity. First, evidence suggests that factory-reared males show similar (Z. cucurbitae; Manoukis and Gayle 2016) or lower (C. capitata; Shelly and Edu 2009) attraction to traps compared to wild males. Thus, if anything, use of factory-reared males is a conservative approach and yields underestimates of trapping sensitivity. In addition, the Carey Group’s implication that the mark-release studies lose credibility because they were not conducted in natural fly habitat is illogical, especially considering that these assessments were conducted in the very areas where the Carey Group claims such flies are now established. What matters is the detection probability in the high-risk, suburban areas of California, not comparable data for the forests of sub-Saharan Africa or Southeast Asia. Finally, the Carey Group’s statement implies that tens of thousands of flies are simply dumped into the field, resulting in dense clusters of highly agitated flies that are more likely to disperse and eventually locate a trap. However, even in a replicated test in which hundreds of thousands of C. capitata were eventually released, only a few hundred flies at most were released at any one point in any given trial by allowing flies to emerge from holding containers of their (Lance 2014). The resulting of flies in the all of the tests that were in detection trapping grids would be considered very to areas where these species are to be established. In the only that the Carey Group has regarding the of data is that the available information with their of sub-detectable we would argue that available estimates of detection system sensitivity would have to be by at least an order of for B. dorsalis two of viable, broadly distributed populations of non-native tephritids to undetected in California. present, we have no to believe that available estimates of detection sensitivity are (Lance 2014). of program that when trapping is a fly detection five to increase in trapping in in the with fruit to the in over half the cases no fly is of the infestation that would trigger an eradication This also counter to the Carey Group that there is level which even a very small population will not be detected (Lance 2014). The Carey Group the operational of eradication for tephritid populations under the for 2016). argue that using a time to a of three generations without detections to eradication is even historically it has and simulations have and 2014). do not regulatory evidence of but only individuals in a and not the of trap detections, as evidence of eradication (Carey 2017). This argument has several major First, trapping is a of detecting flies, how one that a population has individuals without trapping of the the Carey Group (2017) that the requires detection and an analysis of all trap detections in California over the on one trapping are to eradication on the other are to populations with by on the Carey Group the and year that fruit and are by international partners would if they detected or even a of fruit fly the Carey Group international by the of the and to and international agricultural et al. 2014). These international or become for countries that are of the 1994, Even the Carey Group this global regulatory as the on the of and requires and the use of and evidence to that 2012). are by of subject and then and by all that are to the This process that the level of of all are consistently and are and as the et al. 2014), thereby allowing and evidence by fruit and to similar a is by and in international agricultural 2012). Finally, while on a the Carey Group has never any in of how many trap at what would be and for how many generations or years based on their to be to what they call eradication The Carey Group highlighted this case as a of evidence of establishment of a tephritid pest in California. However, several Why were the B. dorsalis the to more than km2 before they be eradicated, even according to the Carey Group, tephritid pest species long post-invasion lag times in population Why were there a number of detection when no oriental fruit flies were The most recent B. three years an generations of the pest which checks of methyl eugenol traps traps serviced biweekly over three oriental fruit fly in the San Diego a given the strong of this to the oriental fruit This is not the traps that were also in the San Diego for months and over the three which represent an those three traps are relocated eight times per a of points are per or points in the San Diego Why has B. dorsalis not been detected on the of the where non-native fruit fly trap are by the in with and the of the of In an extensive medfly was detected in the of that required a to achieve eradication et al. 2015). However, in spite of there has been no medfly or other non-native fruit fly detection on the even San Diego and have into a large The Carey stated that it is that introductions of the oriental fruit fly into San Diego would no flies were detected for the previous The of tephritid in California directly with the of the that in the in the with the of and and they established of and that in the of commercial and thereby increasing the likelihood of introductions of non-native The arrival of insects in and from hundreds of countries can explain the detections in years eradication et al. California has historically been a point to and from it is not a that it has been so to oriental fruit fly and other Bactrocera when compared to other high-risk such as between California and in of numbers of Bactrocera at of from all years pest One of the of the Carey Group is to as an to address fruit flies in a more and However, Carey et al. (2017) and et al. that they have recently in fruit production and as Queensland fruit fly, Bactrocera has invaded and recently become established In the (the of what was the the pest has been with and in (the the had to be a they are now with the by of and of to be to from this up their by the as by the Carey Group, would and the of in California, and other major fruit and The arrival of non-native tephritids be there is a major in at of to increase without and et al. are also no of and eradication and trapping can be further to increase this may not to in the detection and of 2014). agricultural and need to maintain a they have to and trapping that are not but still or their rapid detection so that eradication are not (Lance 2014). The Carey Group that they never the question fruit fly populations remain sub-detectable for and that the question was simply a used by the McInnis Group. However, understanding how polyphagous invasive tephritid pest populations can remain undetected over years under favorable climatic and host conditions is to claims of their establishment in California. It is to expect of the establishment to on the and of these established populations so that the can be This essay is to the of McInnis and many to the and of for and tephritid fruit The also with around the directly with the and