We evaluated mating behaviour of B. dorsalis from central and southern Thailand. All flies were sourced directly from the wild via host-rearing and sent to the FAO-IAEA Insect Pest Control Laboratory (IPCL), Seibersdorf Austria, in March 2010. Collection locations were not privately owned and no endangered or protected species were involved in the study. No specific permits were required for the described field studies or for the import of live material into the IPCL. The central Thailand population was reared from Mangifera indica L. (Anacardiaceae) in Saraburi and sent as a batch of approximately 500 pupae; the southern population was reared from Carica papaya L. (Caricaceae) in Nakhon Si Thammarat and sent as a batch of approximately 200 pupae.

Flies were morphologically examined for external and internal genitalic characters to confirm their identity in accordance with taxonomic descriptions (Drew and Hancock 1994). Professor R.A.I. Drew confirmed the identity of the cultures as B. dorsalis based on pinned material, diagnostic micrographs, and genitalia measurements. Further, material from these colonies were used in subsequent integrative taxonomic studies examining molecular and morphological characters that further confirmed both colonies as B. dorsalis (Krosch et al. 2013, Boykin et al. 2014). Representative voucher samples were preserved as dried (pinned) and wet (> 95% alcohol) material at the IPCL and Queensland University of Technology, Brisbane, Australia.

Adult flies were provided a standard diet of enzymatic yeast hydrolysate and sugar (1:3) together with water ad libitum. Sexually mature flies were exposed to egg-cups dosed with commercial guava juice (Rubricon, Rubricon Products, Middlesex, U.K.) as an oviposition stimulant. Eggs were incubated overnight (25 ± 2 °C, 65% R.H.) on moist filter paper placed on wet sponges in Petri dishes and then transferred to carrot diet (Tanaka et al. 1970) for larval development (27–28 °C, 55% R.H.). Pupae were collected into and sifted from moistened teak sawdust, and transferred to either experimental (20 cm diameter × 27 cm height) or colony cages (50 cm × 50 cm × 50 cm).

The first mating compatibility tests were conducted in June and July of 2010 when Saraburi and Nakhon Si Thammarat colonies were at the F1/F2 laboratory colony generation. Eight replicates were completed, consisting of five using F1 generation flies and three using F2 generation flies. The second series of mating compatibility tests were undertaken one-year later in July 2011 when both cultures had reached F6 (eight replicates completed). Experimental protocols were identical for 2010 and 2011 trials, as outlined below.

Flies were sexed within four days of emergence; this is well before male and female sexual maturation which occurs 15-20 days post emergence based on personal observation (MKS; data not shown) and previous studies (McInnis et al. 1999, Wee and Tan 2000). Flies were maintained under low-stress conditions of 100-200 flies per cylindrical cage (20 cm diameter × 27 cm height). General procedures followed the FAO/IAEA/USDA (2003) Manual for Product and Quality Control. A small dot of coloured water-based paint was applied to the dorsal surface of each fly’s thorax using a soft paint-brush (colours were randomized among tests) for identification of each population. Painting was done at least 48 hours prior to each field cage test to allow paint to dry and flies to become habituated to its presence. Mating tests were undertaken using flies between 20-30 days of age to ensure sexual maturity had been reached by the majority of individuals (> 90% of wild B. dorsalis reach maturity by 24 days (Wong et al. 1989)).

Field cage tests were conducted inside a glasshouse exposed to natural light and maintained at ~25 °C and ~50% R.H. Replicates were undertaken inside one of four partitioned flight cages (2.0 m × 1.6 m × 1.9 m) within the glasshouse, with each cage containing a single, non-fruiting potted Citrus sinensis Osbeck (Rutaceae) tree of 2 m in height with a canopy of ~ 1.1 m in diameter.

Flies were released into the experimental field cage at a 1:2 male:female ratio. As this study was focussed on mating compatibility and not strictly competition, this ratio of males to females (as opposed to 1:1) was used to ameliorate the effect of potentially early-mating males from monopolising all females from one population and thus inflating isolation indices, as per Schutze et al. (2013). By providing twice as many females as males, potentially later starting males still have access to females from both populations. As B. dorsalis mates at dusk (Arakaki et al. 1984), 20 males of each of the two populations were released into a field cage one to two hours prior to sunset for each replicate; 40 females of each of the same two populations were released 30 min later for a total of 40 males and 80 females per replicate. Experimental observations began immediately after females were released. Once couples formed, they were gently coaxed into sequentially numbered plastic vials (3.7 cm diameter × 4.0 cm). The following data were recorded for each pair: male origin; female origin; time of mating; and position (cage or tree). Periodic measurements of temperature (°C) and relative humidity in the cage were also made. Experiments concluded when flies became inactive, which occurred after sunset when light intensity dropped below 10 lux.

Relative percentages of each of the four possible couples (i.e., Saraburi ♂ × Saraburi ♀ [SS], Saraburi ♂ × Nakhon Si Thammarat ♀ [SN], Nakhon Si Thammarat ♂ × Saraburi ♀ [NS], and Nakhon Si Thammarat ♂ × Nakhon Si Thammarat ♀ [NN]) were calculated for each replicate. Proportion data were arcsine transformed prior to subsequent analysis; one-way anova (with Tukey post hoc test where appropriate) was conducted to assess for significant differences among mating combinations within each year; paired t-tests were conducted to assess for significant differences in relative proportions of respective couple combinations across years.

Compatibility was determined using the Index of Sexual Isolation in conjunction with the Male Relative Performance Index and the Female Relative Performance Index (Cayol et al. 1999). Values of ISI may range from +1 (complete positive assortative mating, i.e., males and females only mating with their respective population) to 0 (complete random mating) to -1 (complete negative assortative mating; i.e., all males of one population mating with all females of the opposite population and vice versa). Values of MRPI range from +1 (only males of one population mated; i.e., the first listed in the test) to 0 (males of both populations participated equally in mating) to -1 (only males of the reciprocal population mated; i.e., the second listed in the test). The FRPI is the equivalent of the MRPI but as applied to females.

Ninety-five percent confidence intervals of the used indices (ISI, MRPI, and FRPI) for each of 2010 (F1/F2 flies) and 2011 (F6 flies) were calculated to determine deviations from random mating (ISI = 0) or equal participation by the respective sexes (MRPI & FRPI = 0). Confidence intervals that included zero represent cases of random mating and equal participation between the populations. Heterogeneity chi-square analyses across replicates for each treatment were undertaken to determine if data could be combined prior to further analysis. Following heterogeneity tests, chi-squared tests of independence were applied to determine if males mated predominantly with females of one population over the other.

The mean time to begin mating (mating latency) was estimated by calculating how many minutes had elapsed between the time each couple initiated mating and the time of the first observed mating couple (= time zero) within each particular cage replicate (as per Schutze et al. 2013). Statistical analyses were conducted for each of the tests, with latency data for each of the four possible mating combinations combined prior to one-way ANOVA (with Tukey post hoc test where appropriate) to determine significant differences (α = 0.05) in latency among mating combinations. We conducted a one-way ANOVA (with Tukey post hoc test where appropriate) on arcsine transformed percentage data of couples collected on the tree for each trial to determine if there was a significant difference in couple location (cage or tree).

All values reported represent mean ± 1 s.e. unless otherwise stated.

Our results show that F1/F2 (= ‘wildish’) B. dorsalis from Saraburi and Nakhon Si Thammarat demonstrated significant positive assortative mating: i.e., like was more likely to mate with like than expected by chance. This assortative mating was lost by the 6^th generation, when random mating occurred between the two populations. The change from positive assortative to random mating was most likely due to two factors: latency and relative participation of the sexes. In ‘wildish’ populations Nakhon Si Thammarat males mated sooner than Saraburi males (i.e., their mating latency time was shorter) and Nakhon Si Thammarat females mated more than Saraburi females. The combination of the two attributes led to more Nakhon Si Thammarat × Nakhon Si Thammarat matings. By the 6^th generation, the temporal difference in male latency was lost, as was the increased ‘precociousness’ of the Nakhon Si Thammarat females, leading to random mating between the populations.

Differences in latency in male mating behaviour may be the results of local environmental conditions from where respective populations of B. dorsalis originated. Time of sunset, for example, may be a potential causal factor, considering Nakhon Si Thammarat is located approximately 600 km south of Bangkok and time of sunset (and time of mating) would correspondingly vary. However, despite their geographic distance, time of sunset differs little between these locations across the year: the sun sets approximately 10-12 minutes later in Nakhon Si Thammarat relative to Bangkok in January, yet in July it sets approximately 6 minutes earlier (based on 2014 sunset data; www.sunrise-and-sunset.com). Nevertheless, this slight difference may be sufficient to influence mating latency in early-generation laboratory colony flies. Complexity in circadian rhythm patterns and differences in mating latency between wild and mass-reared colonies have been investigated in other tephritid species, such as the melon fly Bactrocera cucurbitae (Coquillett) (Matsuyama and Kuba 2009); exemplifying how subtle, but significant, differences in the onset of mating behaviour can be readily manipulated by changes in daily light patterns (Miyatake et al. 2002). We cannot rule out other factors besides time of sunset, however, and further work into the driving mechanisms of B. dorsalis mating latency are warranted.

What influenced variation in female mating in our trials remains open to speculation. Drivers of sexual propensity are varied, and may include both intrinsic and extrinsic (e.g., temperature, food, density, and sex ratios) factors (Spiess 1970). Differences in mating propensity may also be affected by body size, as demonstrated in other tephritids such as C. capitata, for which mating frequency was dependant on the relative sizes of males and females (Churchill-Stanland et al. 1986). We did not, however, record additional attributes (e.g., size) of flies used in these trials; therefore, we are unable to account for the contribution of any of these factors towards differences in mating propensity of either sex.

These results pose a conundrum for mating trials. It is generally considered that the use of ‘wildish’ populations (i.e., within one or very few generations of field collected material) is more desirable than using flies that already have been cultured for a long time because laboratory selection may alter key behavioural and physiological traits (Miyatake 1998, 2011, Meats et al. 2004, Gilchrist et al. 2012). However, in our case the use of ‘wildish’ populations led to a result which, when taken alone, was contrary to additional evidence that B. dorsalis constitutes a single biological species within Thailand (Schutze et al. 2012, Krosch et al. 2013, Aketarawong et al. 2014, Boykin et al. 2014). That is, we observed significant assortative mating behaviour driven by behavioural differences which may support a two-species hypothesis; yet, all other forms of data (e.g., molecular, morphological, cytogenetic, and chemoecological) strongly infer con-specificity across Thailand. Nevertheless, subtle differences in mating latency between populations of other Bactrocera species are known, even when there is no suggestion of sibling species. For example, differences in mating latency between populations of B. cucurbitae have been recorded (Miyatake 1998, 2011), but there is no evidence for cryptic taxa within that species (Hu et al. 2008, Virgilio et al. 2010).

Based on this experiment alone we are not in a position to make strong statements about using ‘wildish’ versus older cultures for mating tests in B. dorsalis, but we do highlight that even within a single biological species, local adaptation and drift may lead to subtle but potentially important differences in some aspects of the mating system, as documented in other organisms (Schmitt and Gamble 1990, Berlocher and Feder 2002, Nosil et al. 2005, Cocroft et al. 2010). Perhaps a compromise to ameliorate potential short term behavioural ‘hangovers’ from wild populations is to rear populations for a pre-determined number of generations to standardise colonies in the laboratory, so that any differences found are more likely to reflect underlying genetic causes and not short-term environmental influences. That said, long-term rearing does lead to significant laboratory adaptation, and as demonstrated for some species this bottleneck can occur within the first few generations (Frankham and Loebel 1992, Hoffmann et al. 2001, Gilchrist et al. 2012). The significant reduction in the number of matings occurring on foliage versus cage walls from the F1/F2 flies to the F6 flies is perhaps behavioural evidence of laboratory adaptation, and for within B. dorsalis we suggest conducting compatibility studies on colonies greater than two generations from the wild, but not greater than six.

As a factory quality assurance measure, mating indices serve a valuable function by allowing repeatable and quantifiable measures of the quality of factory flies, thereby forming an effective tool allowing SIT action programme managers to determine if sterile males are fit for purpose to compete with wild males and are compatible with wild females. But there is no doubt that these indices, by focusing exclusively on the ‘end product’ of copulation, may lead some researchers to potentially under appreciate biologically important steps in the courtship process. Where mating is 100% random, or 100% positive-assortative, these prior steps may be less critical for interpreting meaning from the trials. But where the results fall in between these extremes, as has been found in several Bactrocera and Anastrepha studies (Schutze et al. 2013, Vera et al. 2006), the absence of more detailed courtship or mating data makes interpretation difficult.