Endosymbiotic Male-Killing Spiroplasma Affects the Physiological and Behavioral Ecology of Macrocheles-Drosophila Interactions

ABSTRACT

While many arthropod endosymbionts are vertically transmitted, phylogenetic studies reveal repeated introductions of hemolymph-dwelling Spiroplasma into Drosophila. Introductions are often attributed to horizontal transmission via ectoparasite vectors. Here, we test if mites (Macrocheles subbadius) prefer to infect Spiroplasma poulsonii MSRO (Melanogaster sex ratio organism)-infected flies and if MSRO infection impairs fly resistance against secondary mite attack. First, we tested if mites prefer MSRO⁺ or MSRO⁻ flies using pairwise choice tests across fly ages. We then tested whether mite preferences are explained by changes in fly physiology, specifically increased metabolic rate (measured as CO₂ production). We hypothesize that this preference is due in part to MSRO⁺ flies expressing higher metabolic rates. However, our results showed mite preference depended on an interaction between fly age and MSRO status: mites avoided 14-day-old MSRO⁺ flies relative to MSRO⁻ flies (31% infection) but preferred MSRO⁺ flies (64% infection) among 26-day-old flies. Using flowthrough respirometry, we found 14-day-old MSRO⁺ flies had higher CO₂ emissions than MSRO⁻ flies (32% greater), whereas at 26 days old the CO₂ production among MSRO⁺ flies was 20% lower than that of MSRO⁻ flies. Thus, mite preferences for high-metabolic-rate hosts did not explain the infection biases in this study. To assess changes in susceptibility to infection, we measured fly endurance using geotaxis assays. Older flies had lower endurance consistent with fly senescence, and this effect was magnified among MSRO⁺ flies. Given the biological importance of male-killing Spiroplasma, potential changes in the interactions of hosts and potential vectors could impact the ecology and evolution of host species.

IMPORTANCE Male-killing endosymbionts are transmitted from mother to daughter and kill male offspring. Despite these major ecological effects, how these endosymbionts colonize new host species is not always clear. Mites are sometimes hypothesized to transfer these bacteria between hosts/host species. Here, we test if (i) mites prefer to infect flies that harbor Spiroplasma poulisoni MSRO and (ii) flies infected with MSRO are less able to resist mite infection. Our results show that flies infected with MSRO have weaker anti-mite resistance, but the mite preference/aversion for MSRO⁺ flies varied with fly age. Given the fitness and population impacts of male-killing Spiroplasma, changes in fly-mite interactions have implications for the ecology and evolution of these symbioses.

INTRODUCTION

Endosymbiotic bacteria are associated with nearly all forms of eukaryotic life (1–4). In particular, insects and other arthropods commonly harbor bacterial endosymbionts, acting as both primary hosts and vectors (5–8). These arthropod-bacteria symbioses are complex and may be mutualistic, commensal, and/or pathogenic in nature depending on the constituent species and conditions (9–12). Male-killing endosymbionts, frequently of the genus Spiroplasma, can have major impacts on host population growth, structure, and evolution (13–15).

Many bacterial endosymbionts of insects are vertically transmitted. In particular, male-killing Spiroplasma strains, e.g., Spiroplasma poullisoni MSRO (Melanogaster sex ratio organism), are transmitted from mother to daughter and typically kill male offspring before maturity (16–18). Despite the generally close coevolutionary relationships between host insects and their Spiroplasma endosymbionts, phylogenetic research has shown Spiroplasma spp. were introduced into the host genus Drosophila multiple times (19–21). However, the mechanism by which these introductions occurred is not clear.

Infection with ectoparasitic mites is a frequently hypothesized mechanism behind these introductions (22–24). Under laboratory conditions, Macrocheles subbadius can take up male-killing S. poulsonii while feeding on host hemolymph and transmit it intra- and interspecifically to naive fly hosts (25). Circumstantial natural observations also support the hypothesis that mites transfer Spiroplasma horizontally between flies. For example, wild-caught Macrocheles species infecting Drosophila hydei harbor S. poulsonii that is genetically similar to the S. poulsonii found in the fly hosts (21). Macrocheles show consistent attachment behavior, i.e., mites that infect one fly are more likely to attach to additional flies, creating an opportunity for mites to mechanically vector hemolymph-dwelling endosymbionts (26).

The prevalence of Spiroplasma infection among female flies can be low, typically below 50% and often below 10%, across multiple Drosophila-Spiroplasma associations (27, 28). In turn, mites may be unlikely to encounter and infect flies carrying Spiroplasma. However, infection with some strains, especially male-killing lines that also impact female health (such as MSRO), may increase the chance of secondary infection with mite vectors (29). Here, we test three non-mutually-exclusive hypotheses: (i) mites preferentially infect S. poullisoni MSRO-infected flies over uninfected flies, (ii) mite preferences for MSRO⁺ or MSRO⁻ flies are explained by mite preferences for higher-metabolic-rate (MR) hosts, and (iii) MSRO-infected flies are more susceptible to mite attack (measured as the ability to maintain energetically demanding behavioral defenses).

We test these hypotheses using Drosophila melanogaster Oregon-R infected with MSRO. MSRO is male killing in Drosophila melanogaster Oregon-R (17, 29). Transovarial transmission of MSRO occurs during yolk deposition (30), and a bacterial protein (SpAID) lethally disrupts development in male offspring (17, 31).

Bacterial titers as well as the physiological and behavioral effects of MSRO on Drosophila melanogaster Oregon-R have been measured across the fly life span (29). Infected females experience multiple changes due to MSRO infection, including losing lipid/energy stores, reduced climbing ability, impaired microbial immunity, and, ultimately, shorter lifespans (32). These deleterious effects get stronger as flies age and MSRO titers increase (29). Although D. melanogaster Oregon-R flies regularly live >60 days, MSRO⁺ flies are often in poor condition by approximately ∼4 weeks and experience ≥50% mortality by 35 to 40 days (29, 33, 34). MSRO titers in D. melanogaster Oregon-R are low when flies first emerge, high but still increasing at ∼2 weeks old, and plateau by ∼26 days posteclosion (29).

The hemolymph-feeding mite Macrocheles subbadius is a cosmopolitan ectoparasite that infects several insect taxa, including Drosophila (35–37). Mites are known to distinguish between host species as well as between members of the same species based on chemical cues (38–40). M. subbadius can distinguish between fly hosts based on size, body condition, metabolic rate, and chemical cues such as hemolymph (41, 42). We hypothesize that mites would preferentially infects MSRO⁺ hosts, and ostensibly this preference should increase with fly age as MSRO titers increase. Thus, we tested if M. subbadius selectively infects MSRO⁺ flies over MSRO⁻ flies in pairwise choice tests at 2, 14, and 26 days posteclosion.

Previous studies have shown that, when controlling for fly sex, mass, age, and prior infection, M. subbadius preferentially infects flies with higher metabolic rates in pairwise tests (42). We anticipated MSRO infection would increase the metabolic rates of host flies due to energetic requirements of bacterial proliferation and/or fly compensation. Therefore, we also hypothesized that increased metabolic rates among MSRO⁺ flies would explain mite preferences. To test this hypothesis, we conducted flowthrough respirometry measurements on age-matched flies. Bacterial titer increases with fly age, so we also predicted that metabolic differences would be greater in older flies.

Since MSRO is male killing, we also considered whether M. subbadius prefers to infect male or female D. melanogaster. The congeneric mite Macrocheles muscaedomesticae has size-mediated preferences for female D. hydei flies (43). Since D. melanogaster females are larger than males on average, we hypothesize M. subbadius will preferentially infect female D. melanogaster.

Drosophila resistance against M. subbadius is primarily behavioral and limited by host endurance (44–46). Therefore, we hypothesized MSRO infection reduces the capacity of flies to resist mite infection. Furthermore, we anticipated this deleterious effect would increase with fly age as bacterial titer increased (29). Host endurance assays are a proxy measure of resistance that remove the confounding variable of mite preference (46–48). We used geotaxic endurance assays to measure fly endurance at 2, 14, and 26 days posteclosion, and we predicted that endurance would be lower among MSRO⁺ flies than MSRO⁻ flies.

RESULTS

MSRO infection status affects mite preference.

This experiment tested if mites prefer to infect flies harboring MSRO or flies without MSRO at different ages (2, 14, and 26 days old). On average, 41% of Y mazes were successful: 65/182 with 2-day-old flies, 42/124 with 14-day-old flies, and 61/108 with 26-day-old flies. When flies were 2 days old, mites selected the MSRO⁺ fly in 51% (95% confidence interval [CI], 38 to 64%; N = 65) of successful pairwise choice tests (binom.test [49]). When flies were 14 days old, mites were significantly less likely to infect MSRO⁺ flies and only selected the MSRO⁺ fly in 31% (95% CI, 18 to 47%; N = 42) of successful trials. Lastly, when flies were 26 days old, mites were significantly more likely to infect MSRO⁺ flies and selected the MSRO⁺ fly in 64% (95% CI, 51 to 76%; N = 61) of successful trials. There was an age-dependent effect of MSRO on mite likelihood to infect: mites were equally likely to infect the MSRO⁺ and MSRO⁻ flies at 2 days old, less likely to infect MSRO⁺ hosts when flies were 14 days old, and more likely to infect MSRO⁺ hosts once flies were 26 days old (Fig. 1).

FIG 1.

Proportion of mites that infected the MSRO⁺ fly in pairwise Y-maze tests at different fly ages. MSRO⁺ and MSRO⁻ flies were age matched at 2, 14, or 26 days posteclosion. Data labels show sample size (number of Y mazes). Error bars represent 95% confidence intervals (using binom.test).

Mite preference for female flies.

We tested if mites prefer to infect male or female D. melanogaster flies. In 24 of 39 (62%) successful Y mazes, the mite infected the female fly. The confidence interval of this marginally significant preference overlaps 50% (95% CI, 45 to 77%) (binom.test [49]).

MSRO infection affects fly metabolic rate.

This experiment tested whether MSRO infection impacted fly metabolic rate across fly ages. The standardized metabolic rate was modeled using the lmer function with infection status, fly age, their interaction, and mass (and mass-infection and mass-age interactions) as well as random-effects respirometer position and replicate block (a single run of the respirometer) (50). The interaction between fly age and infection status was a significant predictor of metabolic rate (X² = 5.5, P = 0.019). MSRO⁺ and MSRO⁻ flies produced comparable amounts of CO₂ from 2 to 11 days posteclosion, MSRO⁺ flies produced more CO₂ than MSRO⁻ flies at 15 days posteclosion, and then, after 26 days posteclosion, MSRO⁻ flies had higher respiration rates (Fig. 2a). Fly mass was not a significant predictor of metabolic rate (X² = 3.412, P = 0.065), and the correlation between mass and metabolic rate was very low (R² = 0.068; cor.test) (Fig. 2b). Interactions between mass and infection status (X² = 1.6, P = 0.2) and fly age and mass (X² = 3.0, P = 0.09) were not significant.

FIG 2.

(A) Metabolic rates (measured as rate of carbon dioxide production) of MSRO⁺ (dark bars) and MSRO⁻ flies at different fly ages. Cohorts of MSRO⁺ and MSRO⁻ flies were age matched, and CO₂ production was measured using flowthrough respirometry. Data labels show sample size (number of flies). Error bars represent 1 SEM. An asterisk indicates significant difference within an age category (P < 0.05). (B) Metabolic rate versus fly frozen mass. Mass was not a significant predictor of metabolic rate (R² = 0.068), and neither were the mass-infection and mass-age interactions.

There was no significant difference in metabolic rate between MSRO⁺ and MSRO⁻ flies at 2 days old (X² = 0.11, P = 0.74), and there was no significant difference at 6 days (X² = 0.54, P = 0.46) or 11 days (X² = 0.010, P = 0.92). At 15 days old, MSRO⁺ flies had higher metabolic rates (X² = 6.7, P = 0.0097). At 20 days there was no significant difference (X² = 0.55, P = 0.46). At 26 days (X² = 7.4, P = 0.0065) and 34 days, MSRO⁺ flies had significantly lower metabolic rates (X² = 4.4, P = 0.036).

Effect of fly metabolic rate on mite behavior.

In 25 out of 32 (78%) Y mazes, the mite was located in an arm containing a fly. In 17/25 (68%) of trials with a mite choice, the mite was in the arm with the high-metabolic-rate fly. Mites were marginally more likely to be in the arm with the high-metabolic-rate fly (z = 1.76, P = 0.079). The confidence interval of this marginal preference overlapped 50% (95% CI, 46 to 85%; binom.test). Due to the small number of infections, we did not statistically test if more high-metabolic-rate flies were infected. However, in the 6 trials where the mite infected a fly, the mite infected the high-metabolic-rate fly 5 times. This experiment suggests mites show a marginal attraction to higher-metabolic-rate flies and there is little evidence mites prefer to infect low-metabolic-rate flies.

MSRO impacts fly endurance in geotaxis assays.

These experiments tested fly resistance while eliminating the confounding effects of mite proclivity to infect. We recorded the number of cycles before flies were exhausted (c) as well as the time until exhaustion (t). We modeled dependent variables c and t with infection status, fly mass, fly age, infection-age interaction, age-mass interaction, infection-mass interaction, and replicate block as a random effect.

The number of cycles to exhaustion (c) was best described with a negative binomial distribution and was modeled using the glmer.nb function (50). The interaction between fly age and infection status was a significant predictor of c (X² = 10.1, P = 0.0015). Mass was not a significant predictor of c (X² = 1.14, P = 0.29). The interactions between infection status and mass (X² = 1.28, P  =  0.26) and fly age and mass (X² = 1.37, P = 0.24) were also not significant. At 2 days posteclosion, there was no significant difference in c between infected and uninfected flies; MSRO⁻ flies had a mean endurance of 132.0 ± 6.6 (mean ± standard error of the mean [SEM]) cycles, and MSRO⁺ flies had an endurance of 128.0 ± 6.7 cycles (X² = 1.4, P = 0.25). There was a general decrease in number of cycles climbed as flies aged, but this effect was larger among MSRO⁺ flies. At 14 days posteclosion, the MSRO⁺ flies had lower endurance than MSRO⁻ flies, 43.3 ± 4.9 cycles and 68.4 ± 3.4 cycles, respectively (X² > 100, P < 0.0001). At 26 days posteclosion, the MSRO⁺ flies had lower endurance than MSRO⁻ flies as well, 4.6 ± 1.3 cycles and 16.1 ± 1.9 cycles, respectively (X² > 100, P < 0.0001).

The time to exhaustion (t) was modeled with the lmer function (50). The interaction between fly age and infection status was a significant predictor of t (X² = 7.3, P  =  0.0069). Mass was not a significant predictor of t (X² = 0.024, P  =  0.88). The interactions between infection status and mass (X² = 2.748, P  =  0.097) and fly age and mass (X² = 2.12, P  =  0.15) were also not significant. At 2 days posteclosion, t was also comparable between MSRO⁻, 492.4 ± 12.2 s, and MSRO⁺, 514.7 ± 13.9 s, flies (X² = 1.37, P  =  0.24). At 14 days posteclosion, MSRO⁺ flies had substantially lower average t than MSRO⁺ flies: 269.3 ± 28.8 s and 423.3 ± 19.4 s, respectively (X² = 8.2, P  =  0.0041). Similarly, at 26 days posteclosion, MSRO⁺ flies had lower t than MSRO⁺ flies; MSRO⁺ flies had a mean t of 43.2 ± 12.7 s, whereas MSRO⁻ flies had a mean t of 149.8 ± 16.6 s (X² = 15.0, P  =  0.00011). Fly endurance (c and t) generally decreased with age, but the magnitude of decrease was larger among flies harboring MSRO (Fig. 3a and b).

FIG 3.

Endurance of MSRO⁺ and MSRO⁻ flies in negative geotaxis endurance assays. (A) The number of ascents flies made following knockdown. (B) The length of time (seconds) climbed before exhaustion. (C) Mass of flies (milligrams). (D) Time to ascend vial in first cycle, excluding flies that failed to climb (seconds). Data labels show sample size (number of flies). Error bars represent 1 SEM (A to C) or 10th to 90th percentiles (D). An asterisk indicates a significant difference within an age category (P < 0.05).

We also recorded the time the fly took to ascend the vial following the first knockdown (i.e., first cycle time, or f). All flies from both the MSRO⁺ and MSRO⁻ groups climbed the vial at least once at 2 days and 14 days posteclosion. At 26 days posteclosion, 94% of MSRO⁻ flies climbed all the way to the mark, whereas only 47% of MSRO⁺ flies climbed successfully. If a fly failed to ascend the vial once, the first cycle time was arbitrarily set to f  =  11 s for nonparametric analysis (wilcox_test [51]).

The median first cycle time of the MSRO⁺ and MSRO⁻ groups at 2 days posteclosion were nearly indistinguishable at 3.3 s (2.9 to 4.1, 10th to 90th percentile) and 3.3 s (2.6 to 3.9), respectively; this was not a significant difference (wilcox_test, z  =  1.1, P  =  0.25). At 14 days posteclosion, the median first cycle time of MSRO⁺ flies was 4.1 s (3.1 to 6.2) and 6.4 s (3.6 to 8.5) for MSRO⁻ flies (z  =  2.9, P  =  0.0041). At 26 days posteclosion, the median first cycle time was 7.0 (4.9 to 9.3) for MSRO⁻ flies and 11 s (7.9 to 11) for MSRO⁺ flies (z  =  4.9, P < 0.0001). The difference between MSRO⁻ and MSRO⁺ flies at 26 days posteclosion remained significant even when only considering flies that successfully climbed (z  =  3.0, P  =  0.0025), for which the median time was 6.9 s (4.7 to 8.9) and 8.8 s (6.3 to 9.7), respectively. Older flies were slower to ascend the vial on the first cycle, and this was more pronounced among MSRO⁺ flies (Fig. 3d).

Across all fly ages, the mean mass of the MSRO⁺ flies was 1.14 ± 0.01 mg (N  =  95), and the mean mass of the MSRO⁻ flies was 1.10 ± 0.01 mg (N  =  93). This difference was statistically significant (t test, t  =  3.16, P  =  0.015) but relatively small: 4.12%. Among 2-day-old flies there was no significant difference in mass (<1%) between MSRO⁺ and MSRO⁻ flies, 1.07 ± 0.01 mg and 1.06 ± 0.01 mg, respectively (t  =  0.24, P  =  0.81). Among 14-day-old flies there was a significant difference in mass between MSRO⁺ and MSRO⁻ flies, 1.14 ± 0.02 mg and 1.07 ± 0.01 mg, respectively: 6.3% (t  =  3.16, P  =  0.002). Similarly, there was a significant 6.8% difference in mass between 26-day-old MSRO⁺ and MSRO⁻ flies, 1.22 ± 0.02 mg and 1.14 ± 0.01 mg, respectively (t  =  2.73, P  =  0.008). Among the older age categories, MSRO⁺ flies were significantly heavier than MSRO⁻ flies (Fig. 3c).

DISCUSSION

We set out to test three non-mutually-exclusive hypotheses: (i) mites preferentially infect S. poullisoni MSRO⁺ flies over uninfected flies, (ii) mite preferences for MSRO⁺ or MSRO⁻ flies are explained by mite preferences for high-metabolic-rate hosts, and (iii) MSRO-infected flies are more susceptible (measured as the ability to maintain energetically demanding behavioral defenses) to mite attack. The latter hypothesis predicted that flies infected with MSRO would have reduced endurance in negative geotaxis assays (a proxy measure of Drosophila mite resistance [46]). This prediction was borne out in our data (Fig. 3a, b, and d). The MSRO infection-fly age interaction was a significant predictor of endurance; specifically, 14- and 26-day-old MSRO⁺ flies had weaker endurance than MSRO⁻ flies of the same age (Fig. 3a and b). These results show MSRO infection impairs fly endurance, and this deleterious effect increases as flies age and/or MSRO titer increases.

Our geotaxis results are compatible with previous observations that Spiroplasma infection can reduce the percentage of flies climbing in group geotaxis assays at older ages (29, 31). One hundred percent of MSRO⁺ and MSRO⁻ flies in our trials climbed at 2 and 14 days posteclosion, whereas at 26 days posteclosion, 94% of MSRO⁻ flies climbed and only 47% of MSRO⁺ flies successfully climbed. However, by measuring the endurance of individual flies, we detected adverse effects of Spiroplasma infection earlier in the fly life span (Fig. 3a and b). We observed substantially reduced cycles to exhaustion and time to exhaustion among MSRO⁺ flies 2 weeks posteclosion, earlier than studies that only measured the proportion of flies climbing (29, 31, 48). Our results suggest it may be useful in future studies to consider individual organism assays when subtler effects of endosymbionts on fly fitness are important, e.g., during interactions with a secondary parasite.

Hastened senescence among MSRO⁺ flies could explain endurance decreasing faster among MSRO⁺ flies. Shortened fly lifespans, earlier reductions in metabolic rate, and declines in age-specific mobility all suggest MSRO hastens senescence (29, 52) (Fig. 2a and 3a and b). MSRO could influence fly senescence through changes in metabolism, energy reserves, and/or immunity (reviewed in reference 53). Activation of the insect immune system can speed senescence (53, 54). However, MSRO generally evades D. melanogaster immunity (32), suggesting that if MSRO impacts fly senescence it is more likely through effects on metabolism or energy reserves.

We also hypothesized that mites would preferentially attack D. melanogaster infected with MSRO. However, results showed that M. subbadius infection bias depended on fly age: mites were less likely to infect MSRO⁺ 14-day-old flies but were more likely to infect MSRO⁺ flies at 26 days (Fig. 1). Mites did not show a bias toward MSRO⁺ or MSRO⁻ hosts when flies were 2 days old (Fig. 1). Both 14-day-old and 26-day-old D. melanogaster Oregon-R flies have high levels of MSRO, but MSRO titers are higher and plateau by ∼26 days (29). Whether replication-phase or plateau-phase MSRO has a higher potential for horizontal transmission is currently unknown (25). Future research should compare the ability of MSRO collected from hosts of various ages to establish persistent infection following horizontal transmission.

M. subbadius discriminates between potential hosts based on many traits. For example, they prefer larger flies and female flies (55). However, in our preference experiments, we size-matched MSRO⁺ and MSRO⁻ females to within 5% body mass, meaning fly sex and potential mass differences were not driving the preferences observed here (Fig. 1 and 3c). Flies were also restrained, eliminating differences in mobility between groups (Fig. 3a, b, and d). A previous study demonstrated M. subbadius preferentially infects D. nigrospiracula with higher metabolic rates (42). Thus, we hypothesized that fly metabolic rate would explain the preferences of mites for/against MSRO⁺ flies. If metabolic rate explained preferences/aversions for MSRO⁺ flies, we would anticipate MSRO⁺ flies having lower metabolic rates at 2 weeks old and higher metabolic rates at 26 days old relative to MSRO⁻ flies. Contrary to our expectations, MSRO⁺ flies had higher metabolic rates than MSRO⁻ flies 15 days posteclosion and lower metabolic rates than MSRO⁻ flies at 26 and 34 days posteclosion (i.e., the opposite of observed pairwise choices) (Fig. 2a, Table 1).

TABLE 1.

Summary of mite preferences and corresponding fly metabolic rates

Fly age (days)	Mite preference	Higher-MR fly
2	No preference	No difference
14/15	MSRO−	MSRO+
26	MSRO+	MSRO−

The inverted relationship between metabolic rate and mite preference observed here indicates (i) mites preferred to infect low-metabolic-rate flies or (ii) mite preferences for other fly traits overpowered/suppressed the preference for high-metabolic-rate flies (Table 1). However, the mite population used in these experiments is known to prefer hosts with higher metabolic rates (42). This trend is consistent within this line of M. subbadius. It also prefers mated Drosophila females over unmated females, and mated females have higher metabolic rates than unmated females (55–57). Mites had a marginal attraction toward higher-metabolic-rate D. melanogaster. Taken together with prior results, this attraction, albeit weaker than the preference for higher-metabolic-rate D. nigrospiracula, suggests it is unlikely mites were merely selecting flies based on low metabolic rates (42).

It is not currently known what specific cues mites were responding to when infecting MSRO⁺ or MSRO⁻ flies. Mites could conceivably directly sense MSRO-derived compounds and/or fly-derived signals of health. Furthermore, the cues influencing mite infectious behavior may change over the fly life span. Mites may prefer MSRO⁺ flies at 26 days old due to a preference for flies in poor body condition, and MSRO infection appears to speed aging and declining health (41). We expected 14-day-old MSRO⁺ flies to be attractive to mites due to high metabolic rates and relatively high endosymbiont titers. However, mites counterintuitively avoided these flies (Fig. 1). D. melanogaster produces pheromones that indicate physical health that are sensed by conspecifics and potentially ectoparasites (58–60). Since flies may partially compensate for MSRO infection in the short term, they may exhibit normal cues or even exhibit cues that mites find repellent (29). Alternatively, rapidly proliferating MSRO may pose a risk to mites, although the effects of consumed MSRO on mites are not known (25). Regardless of the specific mechanism(s), the fact remains that mites were disproportionately likely to infect either MSRO⁺ or MSRO⁻ hosts depending on fly age.

Mites use contact and olfactory chemoreception to find and distinguish between potential hosts (59, 61, 62). Poecilochirus species choose larger host beetles, potentially based on chemical cues (63). Dermanyssus gallinae is attracted to volatile cues associated with the aged feathers, but not fresh feathers, of host birds (64). Another mesostigmatid mite, Varroa destructor, discriminates between nurse and forager bees based on differences in cuticular hydrocarbons (65). MSRO has significant impacts on host physiology, metabolism, and body composition (29) (Fig. 2a). All these changes could have knock-on effects on host cues (66–68). Since mites in our preference experiments could contact flies before infection, it is possible that mites were using nonvolatile hydrocarbons on the surface of flies or volatile kairomones to discriminate between hosts (69). At present, it is not known how MSRO, directly or through knock-on effects on fly condition/physiology, impacts the composition of cuticular waxes or volatiles. Because mite preferences changed across the fly life span, it may be fruitful to examine these cues in MSRO⁺ and MSRO⁻ flies at the ages identified here (Fig. 1).

Previous studies showed Macrocheles muscadomesticae mites preferentially infect D. hydei females over conspecific males in a size-mediated manner (43). Drosophila melanogaster flies are also sexually dimorphic in body mass, with female flies also being larger on average than male flies (70, 71). However, the average size of D. melanogaster is substantially smaller than that of D. hydei (approximately half or less by mass), and resultingly the absolute size difference between male and female D. melanogaster flies is smaller (43). Smaller absolute size differences may explain the weaker preference for females observed here.

The impact of MSRO on fly metabolic rate varied over the fly life span. At 14 days old, MSRO⁺ flies had higher metabolic rates, whereas at ≥26 days MSRO⁺ flies had lower metabolic rates (Fig. 2a). We did not detect a significant effect of frozen mass on female fly metabolic rate (Fig. 2b), which is consistent with many lines of lab D. melanogaster (72), nor did we detect a significant interaction between MSRO infection status and mass. One potential explanation for why older MSRO⁺ flies had lower metabolic rates than age-matched MSRO⁻ flies is hastened senescence, since D. melanogaster metabolic rates tend to decrease with age (73–75). Likewise, mobility also decreases with fly age (76, 77) (Fig. 3). However, rapid senescence alone does not explain elevated metabolic rates among 14-day-old MSRO⁺ flies. Fourteen days old corresponds to mid-late log growth and may correspond to periods of high endosymbiont energy use (29). Alternatively, flies may compensate for infection, e.g., by mobilizing stored energy or increasing feeding, leading to higher metabolic rates.

Compensating for early proliferation may preserve fly health during periods of high fly fecundity, before reproduction significantly decreases ∼3 weeks posteclosion (78, 79). MSRO⁺ D. melanogaster flies lay eggs slightly sooner than MSRO⁻ flies, suggesting they may partially compensate for early death/poor health with earlier ovipositing (29). Organisms have a finite capacity to compensate for infection, and compensation could even come with a trade-off of earlier senescence due to effects of metabolism on Drosophila aging (80). Future studies could test if MSRO⁺ flies engage in compensatory feeding across the fly life span. Immuno-mutant and metabolic-mutant D. melanogaster may also provide a method evaluating the relative contributions of the fly and bacterium to metabolic changes observed here (81, 82).

Endosymbiotic bacteria vary substantially in their fitness level effects and impacts on host insect health. MSRO has substantial negative effects on fly life span and body condition (29) (Fig. 3a and b). However, other Spiroplasma species have smaller effects or even are beneficial, for example contributing to parasitoid resistance (83–85). Spiroplasma HapI has little to no effect on fly survival but protects D. hydei hosts against parasitoid attack (86, 87). Other endosymbionts show similar variability. Wolbachia pipientis wMelPop is pathogenic and reduces fly longevity (88, 89), whereas other Wolbachia strains have relatively mild effects, such as minor changes in mobility, or are beneficial (85, 90). Additionally, the impacts of endosymbionts on hosts can be environment dependent (91, 92). Thus, it is important for future research to compare the impacts of different endosymbiont bacteria on host-ectoparasite interactions across environments. It may be possible that some endosymbionts improve ectoparasite resistance (90).

In this study, MSRO⁺ flies had reduced endurance (a proxy of mite resistance), but mites preferred MSRO⁺ or MSRO⁻ flies at different fly ages. Further research is required to determine if differential susceptibility leads to mites disproportionately infecting MSRO⁺ flies on a population level and should test if wild-caught flies harboring male-killing Spiroplasma are more likely to be infected with natural ectoparasites. Given their fitness and population effects (14, 15, 93), the impacts of male-killing Spiroplasma on the interactions between hosts and potential vectors have implications for the ecology and evolution of these symbioses.

MATERIALS AND METHODS

Cultures.

Drosophila melanogaster Meigen (Oregon-R) (Diptera: Drosophilidae) cultures, both MSRO⁺ and MSRO⁻ lines, were maintained on agar media at 25°C on a 12-h:12-h light-dark cycle. The culture temperature ensured reliable vertical transmission of MSRO (83, 94). Fly fecundity is highest ∼1 week posteclosion and decreases with age before dropping substantially ∼3 to 3.5 weeks posteclosion (78, 79). MSRO infection does not significantly affect the number of eggs laid by D. melanogaster but may shift flies toward earlier reproduction (29). Since MSRO kills males before the adult stage, 3 to 5 males from MSRO⁻ stocks were added to each MSRO⁺ vial for mating. Age-matched female flies for experiments were collected within 24 h of eclosion and stored in vials with 3 to 5 males (total population density, 23 to 25 flies/vial). When flies were transferred to new vials, males were replaced as necessary to ensure all experimental flies were mated. To confirm this method led to reliable mating, we conducted an additional test where we mated female flies, as in the experiments, and monitored them for egg production. We informally tracked survival of flies as they aged. When flies were aged beyond ∼26 days, mortality sometimes required reducing vial density, so MSRO⁺ and MSRO⁻ vials were split as needed to keep fly populations at equal densities. Approximately 95% of flies lived to ∼3 weeks posteclosion regardless of MSRO infection status. However, ∼3.5 to 4 weeks posteclosion, MSRO⁺ fly survival dropped precipitously, and only 40 to 50% of MSRO⁺ flies lived to 34 days, while >80% of MSRO⁻ flies did (personal observation). Fly survival was overall consistent with previously reported survival (29).

The MSRO⁺ culture was initially generated by microinjecting the MSRO⁻ line with hemolymph from infected flies. Lines have been cultured since at least 2015, eliminating potential direct and maternal effects of initial MSRO establishment. Periodically, we inspected MSRO⁺ vials for the presence of male flies (examining all the offspring of a vial), which would indicate MSRO was lost or contamination occurred. Any MSRO⁺ vials that produced male offspring were destroyed, and all MSRO⁺ flies used in experiments came from vials that produced only female offspring. Previous research shows MSRO titer is low in 2-day-old Drosophila melanogaster Oregon-R, high but still in log growth among 14-day-old flies, and highest but plateaued by ∼26 days posteclosion (29). Thus, preference and endurance experiments were conducted with 2-day-old, 14-day-old, and 26-day-old flies.

Macrocheles subbadius (Berlese) (Mesostigmata: Macrochilidae) cultures were founded from 200 to 300 adult female mites collected in the Sonoran desert from wild flies (Arizona) (35) and were maintained on a 2:1 mix of wheat bran to wood chips. Free-living nematodes (Rhabditida) were cocultured as a food source for mites, and nutritional yeast was added to support nematode and mite growth. Adult female mites were collected for experiments using Berlese funnels. These experimental mites were held overnight on moist plaster of Paris prior to experiments, as M. subbadius is more likely to infect flies if they are held without food (41).

Effect of fly MSRO infection on mite preference.

Y maze experiments tested if mites have a preference for MSRO⁺ or MSRO⁻ flies at 2, 14, or 26 days posteclosion. We made size-matched pairs, within <5% difference in body mass, of adult female flies (1 MSRO⁺ and 1 MSRO⁻ fly) to account for the preference of M. subbadius for larger flies (55). Experimental flies were glued to a small piece of cotton (Elmer’s rubber cement) to eliminate differences in resistance/mobility (43). Size-matched pairs were transferred to the ends of Y-shaped mazes (Y polypropylene connectors; Fisherbrand Tubing), alternating MSRO⁺ and MSRO⁻ on the left/right, and then a single adult female mite was introduced to each maze and the ends were sealed with cotton. The Y mazes were transferred to an opaque box to exclude light, as mites are more likely to infect in the dark (personal observation). Y mazes were left undisturbed for 2 h at ambient temperature and then scored. A Y-maze trial was considered successful if the mite infected one of the two available flies, and in successful trials we recorded if the MSRO⁺ or MSRO⁻ fly was infected. We tested if mites infected MSRO⁺ and MSRO⁻ flies in each age category at the same rate using binomial tests (binom.test function [49]).

Effect of fly sex on mite preference.

The Y-maze preference experiment tested if M. subbadius preferentially infects male or female D. melanogaster. Y-maze trials were conducted as described for the MSRO preference experiments, except flies were not size matched (allowing natural sexual dimorphisms in mass to persist). As MSRO is male killing, both the male and female flies came from the MSRO⁻ line. This experiment was also analyzed with a binomial test (binom.test [49]).

Effect of MSRO infection on fly metabolic rate.

This experiment tested if MSRO impacts the metabolic rates of infected flies and, by extension, if mite preferences for MSRO⁺ or MSRO⁻ flies are potentially mediated by mite preferences to infect flies with high metabolic rates (42). In this experiment, we measured a wider range of ages, primarily to test if metabolic rate at the plateau phase of MSRO remains constant. We measured the metabolic rates of age-matched MSRO⁺ and MSRO⁻ female fly cohorts at 2, 6, 11, 15, 20, 26, and 34 days posteclosion using flowthrough respirometry. Flies were frozen postassay, and mass was weighed when possible. Metabolic rate was measured as the rate of carbon dioxide production (microliters per minute) (95). A MAVEn-FT unit (Sable Systems International, NV, USA) was used to set the flow rate at 30 ml/min and to control which of 15 respirometer positions was active (up to 14 containing flies and 1 empty chamber). MSRO⁺ and MSRO⁻ flies were placed in alternating respirometer positions to reduce the effects of acclimatization/respirometer stress. An Li-7000 CO₂ sensor (Li-Cor Environmental, NE, USA) was then used to measure CO₂ in air that had flown through chambers that contained MSRO⁺ flies, MSRO⁻ flies, or a baseline (empty) chamber. To improve sensitivity, incoming air was purged of CO₂ and water vapor (FT-IR purge gas generator 75-45; Parker Canada Division, Milton, Ontario, Canada), and excurrent gas was purged of water vapor again using a magnesium perchlorate column. Between uses, the glass respirometry chambers were cleaned with detergent, sterilized with 70% ethanol, and rinsed with distilled water. Glass chambers were checked periodically for damage. Real-time flow rates were examined to detect leaks, and flies where there was evidence of leakage (either unexpectedly low flow or erratic flow) were removed prior to analysis. Data were analyzed using the Expedata software (Sable Systems International, NV, USA), and the real-time flow rate reported by the MAVEn-Ft was used to calculate the mean CO₂ production rate for each fly (calculation in reference 96). Fly metabolic rate was standardized before modeling: (X_MR − μ_MR)/σ_MR.

Standardized metabolic rate was modeled using linear mixed-effect models (lmer function [50]), with fly infection status, fly age, the infection-age interaction, and mass as fixed effects and replicate block (a single run of the respirometer) and respirometer position as random effects. The mass X infection and mass X fly age interactions were also included. Starting with the least significant fixed effect, independent fixed variables were removed from the model and the original and new models were compared with the R function anova (old model, new model) (49). Model reduction was also performed for each fly age with the same fixed and random effects.

Effect of fly metabolic rate on mite behavior.

In the experiment on the effect of fly metabolic rate on mite behavior, we considered if mites are attracted to and/or prefer to infect D. melanogaster females with higher metabolic rates over flies with lower metabolic rates. We measured the metabolic rates of MSRO⁻ females (2 to 5 days old) using respirometry (as described above, minus the empty chamber). We paired the highest and lowest (approximately the top and bottom thirds) metabolic rate flies, tethered them to cotton, and placed them in the ends of a Y maze (minimum MR difference, 31%). A single adult female mite was introduced to the Y maze. The Y mazes were covered for 1 h, after which we recorded (i) the final position of the mite (high-MR arm, low-MR arm, or the arm with no fly) and (ii) which fly (if any) was infected. We used the glm function with a binomial distribution to test if mites were attracted to the high- or low-MR fly.

Effect of MSRO infection on fly endurance in geotaxic assays.

We used negative geotaxis endurance assays to measure the resistance capacity of MSRO⁺ and MSRO⁻ flies. Behavioral defense acts as the primary line of resistance against mite infection and is limited by fly endurance. As such, the ability to maintain energetically demanding activity is a proxy measure of resistance that eliminates the effects of mite preferences (46). We marked empty, transparent vials at 5 cm above the base and then aspirated a single MSRO⁺ or MSRO⁻ fly into a vial and allowed 15 min for acclimation. At the outset of the experiment (t  =  0), the experimenter tapped the vial, causing the fly to fall to the bottom of the vial and inducing the fly to climb. This knockdown procedure was repeated whenever the fly reached the 5-cm line. A fly was considered exhausted if it failed to ascend the vial after 10 s.

We recorded two metrics of endurance: the number of knockdowns before the fly was exhausted (cycles, c) and the time until exhaustion (seconds, t); if a fly failed to climb following the initial knockdown, both values were recorded as 0. Time until exhaustion and number of cycles until infection were modeled separately using linear mixed-effect models (lmer, glmer.nb [50]), with fixed effects of infection status (MSRO⁺ or MSRO⁻), fly age, and fly mass as well as replicate block (experiment date) as a random effect. The interactions between infection status and mass, fly age and mass, and infection status and fly age were also included in the initial models. Starting with the least significant fixed effect, independent variables were removed from the model and the original and new models were compared with the R function anova (old model, new model). Model reduction was also performed for each fly age with the same fixed and random effects.

The time it took for the fly to climb to the mark after the first knockdown was also recorded (first cycle time, f). If a fly never ascended the vial, f was set to an arbitrarily high value for nonparametric analysis (f = 11 s). We examined the first cycle time using nonparametric tests (wilcox_test function [51]).

Data availability.

Data are available at Dryad (https://doi.org/10.5061/dryad.t1g1jwt3f).