Mating frequency and resource allocation to male and female function in the simultaneous hermaphrodite land snail Arianta arbustorum

Authors

  • Locher,

    1. Department of Integrative Biology, Section of Conservation Biology (NLU), University of Basel, St. Johanns-Vorstadt 10, CH-4056 Basel, Switzerland
    Search for more papers by this author
  • Baur

    1. Department of Integrative Biology, Section of Conservation Biology (NLU), University of Basel, St. Johanns-Vorstadt 10, CH-4056 Basel, Switzerland
    Search for more papers by this author

Bruno Baur Department of Integrative Biology, Section of Conservation Biology (NLU), University of Basel, St. Johanns-Vorstadt 10, CH-4056 Basel, Switzerland. Tel.: +41 61 267 0829; fax: +41 61 267 0832; e-mail: bruno.baur@unibas.ch

Abstract

Sex allocation theory predicts that mating frequency and long-term sperm storage affect the relative allocation to male and female function in simultaneous hermaphrodites. We examined the effect of mating frequency on male and female reproductive output (number of sperm delivered and eggs deposited) and on the resources allocated to the male and female function (dry mass, nitrogen and carbon contents of spermatophores and eggs) in individuals of the simultaneous hermaphrodite land snail Arianta arbustorum. Similar numbers of sperm were delivered in successive copulations. Consequently, the total number of sperm transferred increased with increasing number of copulations. In contrast, the total number of eggs produced was not influenced by the number of copulations. Energy allocation to gamete production expressed as dry mass, nitrogen or carbon content was highly female-biased (>95% in all estimates). With increasing number of copulations the relative nitrogen allocation to the male function increased from 1.7% (one copulation) to 4.7% (three copulations), but the overall reproductive allocation remained highly female-biased. At the individual level, we did not find any trade-off between male and female reproductive function. In contrast, there was a significant positive correlation between the resources allocated to the male and female function. Snails that delivered many sperm also produced a large number of eggs. This finding contradicts current theory of sex allocation in simultaneous hermaphrodites.

Introduction

Hermaphroditic plants and animals can potentially maximize reproductive success through a wide variety of different strategies. A key observation for testing sex allocation theory in simultaneous hermaphrodites is the proportion of resources devoted to male vs. female function ( Charnov, 1982). The specific allocation strategy followed by a hermaphrodite may affect the extent of sexual selection and mating behaviour ( Michiels, 1998). Most models of sex allocation are based on the concept of male and female gain curves (the relationship between relative investment in either male or female gamete production and resulting reproductive success; Charnov et al., 1976 ; Charnov, 1979, 1982). These models received substantial empirical support (e.g. in coral reef fishes [ Fischer, 1980, 1981, 1984; Fischer & Petersen, 1987; Petersen, 1987, 1991], a polychaete worm [ Sella, 1985, 1990, 1991] and in a barnacle [ Raimondi & Martin, 1991]). Most of these hermaphrodites have external fertilization. A variety of hermaphroditic invertebrates, however, have some form of copulation, sperm storage and internal fertilization ( Michiels, 1998).

More recently, sex allocation models for outbreeding hermaphrodites with internal fertilization and sperm storage have been developed ( Charnov, 1996; Greeff & Michiels, 1999). These models consider how various aspects of sperm competition, such as mating frequency, sperm digestion and different mechanisms of sperm displacement, affect sex allocation in simultaneous hermaphrodites. The models predict that a reduced mating rate leads to a reduction in resources allocated to the male function ( Charnov, 1996; Greeff & Michiels, 1999), while sperm digestion leads to an increase in allocation to the male function ( Greeff & Michiels, 1999). At present, there are no data available on relative male/female allocation for internally fertilizing simultaneous hermaphrodites with different mating frequencies.

The research presented here explores sex allocation in Arianta arbustorum, a simultaneous hermaphrodite land snail with internal fertilization, sperm storage and sperm digestion. We designed an experiment to investigate (1) whether snails alter the relative reproductive allocation to the male function with increasing number of copulations, and (2) whether the energy spent on spermatophores and sperm is traded off against energy expended on egg production. For this purpose we examined the number of sperm delivered and eggs produced by snails that mated once, twice or three times within one reproductive season. As measures of resource allocation we assessed the dry mass, nitrogen and carbon content of spermatophores and eggs produced.

Methods

Study animals

Arianta arbustorum is a simultaneous hermaphrodite land snail, which is common in moist habitats of north-western and central Europe. The snail has determinate growth (shell breadth of adults 17–22 mm). Individuals become sexually mature at 2–4 years, and adults live another 3–4 years (maximum 14 years; Baur & Raboud, 1988). In the field, snails deposit 1–3 egg batches, consisting of 20–50 eggs, per reproductive season ( Baur & Raboud, 1988; Baur, 1990). Breeding experiments showed that 27% of virgin snails prevented from mating produced a few hatchlings by self-fertilization in the second and third year of isolation ( Chen, 1993). The reproductive success of selfing individuals, however, was less than 2% of that of mated snails, suggesting high costs for selfing ( Chen, 1994).

Mating in A. arbustorum includes elaborate courtship behaviour with optional dart shooting (i.e. the pushing of a calcareous dart into the mating partner’s body), and lasts 2–18 h ( Hofmann, 1923; Baur, 1992a). Copulation is reciprocal; after intromission each snail transfers simultaneously one spermatophore ( Haase & Baur, 1995). The spermatophore is formed and filled with sperm during copulation ( Hofmann, 1923). It has a distinctive form consisting of a head, a body (sperm container with 800 000–4000 000 spermatozoa) and a tail 2–3 cm long ( Baur et al., 1998 ). The snails mate repeatedly in the course of a reproductive season, and fertile sperm can be stored for more than 1 year ( Baur, 1988). Mating was found to be random with respect to shell size and different degrees of relatedness ( Baur, 1992a; Baur & Baur, 1997). Individuals need 8 days to completely replenish their sperm reserves after a successful copulation ( Locher & Baur, 1999). There is a high degree of reciprocity in spermatophore transfer, but no sperm trading with respect to sperm number occurs ( Baur et al., 1998 ).

Paternity analysis in broods of wild-caught A. arbustorum showed a high frequency of multiple insemination ( Baur, 1994). A controlled laboratory experiment showed that one successful copulation per reproductive season is sufficient to fertilize all the eggs produced by an individual ( Chen & Baur, 1993). However, there is a probability of 5–8% that a copulation will not lead to fertilization of eggs (no sperm transfer or transfer of unfertile sperm; Chen & Baur, 1993).

General methods

To obtain virgin snails we collected subadult individuals that had not yet completed shell growth from an embankment along a track in a subalpine forest near Gurnigelbad, 30 km south of Bern, Switzerland (46°45′N, 7°28′E) at an altitude of 1250 m. The snails were kept isolated in transparent beakers (8 cm deep, 6.5 cm in diameter) lined with moist soil (approximately 4 cm) at 19 °C and on a light:dark cycle of 18:6 h for 4 weeks. During this period, subadult individuals reached sexual maturity as indicated by the formation of a flanged lip at the shell aperture. We cleaned the beakers twice per week, and provided fresh lettuce ad libitum as food. The snails were marked individually with numbers written on their shells with a waterproof felt-tipped pen on a spot of correction fluid (Tipp-Ex). The animals showed no visible reaction to the marking procedure.

Mating trials were performed outdoors to expose snails to natural temperature and light conditions. Two randomly chosen active snails (individuals with an extended soft body and everted tentacles) were allowed to copulate in a transparent plastic container, measuring 14 × 10 × 7 cm, whose bottom was covered with moistened paper towelling to maintain activity. Mating trials were initiated in the evening and ran during 14 nights in June and July. The period between the end of May and middle of July is the time of maximum mating activity in subalpine populations of A. arbustorum.

We observed the snails’ mating behaviour at intervals of 30 min (at night using a torch) following the method described in Baur (1992a) and Baur et al. (1998 ). Records included courtship duration (time interval from courtship initiation to copulation) and copulation duration. Observation sessions were terminated either after successful copulation or after 8 h if no snail initiated courtship behaviour. Snails that did not mate were tested again 3 days later with a new partner. Between two trials, unmated snails were kept isolated as described above.

After copulation, one mating partner (hereafter referred to as sperm donor) was kept isolated in a transparent plastic beaker lined with moist soil (as described above). The other mating partner (hereafter referred to as sperm recipient) was frozen immediately after copulation. Sperm donors were assigned to three groups. The snails of one group were allowed to copulate once, those of the second group twice and those of the third group three times. All matings occurred within a period of 51 days, but the interval between two matings was always larger than 7 days. Snails from the second and third group did not differ in the interval between two successive copulations (t=1.18, d.f.=27, P=0.25).

All sperm recipients were virgin individuals. To obtain the spermatophore we dissected out the female reproductive duct of the recipient. We measured the length (L) and width (W) of the sperm-containing part of each spermatophore to the nearest 0.1 mm using a dissecting microscope. Spermatophore size (in mm3) was approximated, by the formula (πLW2/4), assuming a cylindrical volume. Spermatophores were kept singly in Eppendorf tubes at –30 °C until required.

The beakers of sperm donors were checked twice per week for eggs. The eggs of each batch were collected, counted and kept in a plastic dish (6.5 mm in diameter) lined with moist paper towelling at 19 °C to determine hatching success. Newly hatched snails were separated from remaining unhatched eggs to prevent egg cannibalism ( Baur, 1992b). In all treatment groups eggs were collected over a period of 75 days following the first (or single) copulation. The length of this period corresponds to one reproductive season of A. arbustorum living in the wild ( Baur & Raboud, 1988; Baur, 1990). In the present study most snails ceased to deposit eggs after 60 days.

To assess any effect of the size of the sperm donor on the number of sperm transferred and number of eggs produced, we measured the size (shell breadth and height) of each mating snail to the nearest 0.1 mm using vernier callipers and calculated the shell volume using the formula: shell volume=0.312 × [(breadth)2 × height] – 0.038 (measurements in mm; B. Baur, unpublished data). Shell volume is a more reliable measurement of snail size than weight, because weight depends on the state of hydration and thus is highly variable in terrestrial gastropods.

Sperm counting procedure

We evaluated the number of sperm that an individual delivered by counting the number of sperm in the spermatophore transferred. This procedure is described in detail in Locher & Baur (1997).

The spermatophore of A. arbustorum consists of a hardened secretion which encapsulates the spermatozoa ( Hofmann, 1923). We mechanically disrupted the spermatophore in 200 μL PBS-buffer (138.6 m M NaCl, 2.7 m M KCl, 8.1 m M Na2HPO4 × 2H2O and 1.5 m M KH2PO4) using a pair of microscissors. The sperm suspension was homogenized with a set of Gilson pipettes for 5–15 min. To count the sperm, we stained the homogenate for 1–3 h with an equal volume of a gallocyanin–chromium complex which stains the DNA in the head of the spermatozoa. If spermatozoa still occurred in clusters, we treated the sample overnight with a sonicator (35 kHz). Two subsamples of known volume of the sperm suspension were diluted 1:3 with PBS-buffer and transferred to a Bürker–Türk counting chamber. This counting chamber consists of 16 cells each with a volume of 25 nL. We counted all sperm heads in randomly chosen cells until the total number of sperm heads exceeded 400 and used the average of two subsamples to calculate the total number of sperm in a spermatophore.

Estimate of reproductive allocation

Three measures of reproductive allocation to male and female function were used: (1) the dry mass of spermatophores filled with spermatozoa and that of the eggs produced, (2) the nitrogen content of spermatophores and eggs, and (3) the carbon content of spermatophores and eggs.

Eighteen spermatophores were obtained from copulating A. arbustorum that were kept as described above, but were not used in the main experiment. The volume of the sperm container was assessed as described. Spermatophores were dried for 72 h at 60 °C and their dry mass was recorded to the nearest 0.01 mg. The nitrogen and carbon concentrations of each spermatophore were determined using a CHN analyser (LECO CHN-900, LECO Instruments GmbH, Munich, Germany). Data on dry mass and nitrogen and carbon concentrations of 264 eggs from 33 batches were obtained from Baur & Baur (1998).

Data analyses

We determined the relationship between the dry mass of spermatophores and sperm (Y in mg) and the size of the spermatophores (X=volume of the sperm container in mm3)

inline image

(r2=0.53, d.f.=16, P=0.0007). This relationship was used to calculate the dry mass of the spermatophores produced by single snails in the main experiment. Nitrogen and carbon content of spermatophores were obtained by multiplying the spermatophore dry mass with the average nitrogen and carbon concentrations, respectively. Similarly, the nitrogen and carbon contents of all eggs produced by an individual were obtained by multiplying egg number with mean dry mass of an egg and the corresponding nitrogen and carbon concentrations.

The SAS program package ( SAS Institute, 1991) was used for statistical analyses. Means ± 1SE are given unless otherwise stated. Analysis of variance ( ANOVA) was used to examine differences in male and female reproductive output among groups of snails with different numbers of copulations. Frequency data (hatching success) were arcsin-transformed. Kruskal–Wallis test was applied for courtship and copulation duration. The StatView program package (Version 5.0, Abacus Concepts, 1998) was used for post hoc power tests, defined as P*=1 – β, where β is the probability of falsely accepting H0 when H1 is true. Hence, for nonsiginificant results post hoc power tests for one-way ANOVA at the α=0.05 level were performed. Analysis of covariance ( ANCOVA, type III model) was used to examine differences in slope and intercept of the regression lines between the relative allocation to the male and female function among snails with different numbers of copulations.

Results

Male and female reproductive output

Snails from the three treatment groups did not differ in shell volume (Table 1). Thus, any possible between-group differences in reproductive output may not be a result of differences in snail size.

Table 1.   Shell size and male and female reproductive output of A. arbustorum that mated once, twice or three times within one reproductive season. Mean values ±ISE are presented. n indicates the number of individuals. F- and P-values resulted from one-way ANOVA. Thumbnail image of

The total number of sperm delivered increased from 3098 000 (SE=348 000, n=15) in snails that copulated once to 5001 000 (350 000, n=23) in snails that copulated twice, and to 6849 000 (528 000, n=6) in individuals with three copulations. Considering the number of sperm delivered per copulation, however, there was no difference between snails with different numbers of copulations (Table 1). Snails from different treatment groups also transferred spermatophores of similar size (Table 1). Female reproductive output was not influenced by the number of copulations. Snails that copulated once, twice or three times produced the same number of eggs within one reproductive season (Table 1). Furthermore, hatching success of eggs did not differ between snails with one, two or three copulations.

Considering snails from all groups, there was a positive correlation between the total number of sperm delivered and number of eggs produced (log-transformed data: r=0.31, n=44, P=0.037). This indicates that individuals that delivered many sperm in general also produced a large number of eggs.

Duration of courtship and copulation

The duration of courtship and copulation can be considered as a measure of the energy invested in mating behaviour. Snails from the three groups differed neither in courtship duration nor in copulation duration (median courtship time: 160 min, range 60–360 min, Kruskal–Wallis test, P=0.24; median copulation time: 106 min, range 55–170 min, Kruskal–Wallis test, P=0.16). Furthermore, snails from the three groups did not differ in number of unsuccessful mating trials per successful copulation (mean: 0.78 unsuccessful trials; Kruskal–Wallis test, P=0.84). Considering single snails, the duration of both courtship and copulation did not differ between successive matings of the same individual (sign-test, in all cases P > 0.2). This indicates that the energy invested in mating behaviour does not change in successive copulations.

Dry mass and N- and C-content of spermatophores and eggs

Dry mass of spermatophores delivered averaged 0.91 mg (SE=0.06, n=18, range: 0.44–1.43 mg). Spermatophores had a nitrogen concentration of 12.05% (0.10, n=18). This indicates that a spermatophore on average contained 0.11 mg nitrogen. Correspondingly, the carbon concentration of a spermatophore was 50.51% (0.16, n=18), resulting in an average carbon content of 0.46 mg.

Dry mass of single eggs averaged 2.14 ± 0.17 mg (grand mean of 33 batches; Baur & Baur, 1998). The nitrogen concentration of single eggs averaged 4.02% (0.08) indicating a nitrogen content of 0.086 mg per egg. The carbon concentration of single eggs averaged 32.05% (0.17) indicating a carbon content of 0.686 mg per egg.

Proportion of resource devoted to male vs. female function

Table 2 presents the percentage of dry mass, nitrogen and carbon allocated to the male function. The relative allocation to the male function increased with increasing number of copulations. However, independently of the measure chosen, reproductive allocation was highly female-biased. In none of the measures used did the average resources devoted to the male function reach 5% of the total allocation in snails that copulated three times. Considering individual snails, the maximum nitrogen allocation to the male function was 13.35% in a snail that copulated three times. Thus, only a minor part of the resources available for reproduction were devoted to the male function. Furthermore, snail size did not affect the relative reproductive allocation to male or female function (in both cases r=0.15, n=44, P=0.32).

Table 2.   Relative allocation to the male reproductive function in individuals of A. arbustorum that mated once, twice or three times within one reproductive season. The relative allocation is expressed as percentage of the total dry mass, amount of nitrogen and carbon devoted to the male function. Values in italics refer to the relative allocation per copulation. Figures in square brackets indicate ranges. Notations as in Table 1. F- and P-values resulted from one-way ANOVA. Thumbnail image of

Considering snails from all groups, there was a positive relationship between the total dry mass of all spermatophores delivered and that of all eggs produced during the experiment ( Fig. 1; r2=0.70, n=44, P < 0.0001). Analysis of covariance revealed that the intercepts of the regression lines differed among groups of snails with different numbers of copulations (Table 3). Furthermore, the ANCOVA had a significant interaction term, indicating differing slopes among groups. When snails with three copulations (n=6) were dropped from the ANCOVA, the interaction term was no longer significant (P=0.47), indicating that snails with one and two copulations did not differ in slope. We dropped the nonsignificant interaction term from the model (cf. Scheiner & Gurevitch, 1993) and repeated the ANCOVA. The results showed that the intercepts of the regression lines differed between the two groups of snails (Table 3). Further analyses showed that the difference in the slope of the regression lines between snails with three copulations and those with one or two copulations is entirely due to snail #575 (indicated by an arrow in Fig. 1).

Figure 1.

 Relationship between the relative reproductive allocation to the male and female function, expressed as total dry mass of all spermatophores and all eggs produced in A. arbustorum. Circles: snails with one copulation, triangles: two copulations, squares: three copulations. Statistics are in Table 3. The arrow indicates snail #575.

Table 3.   Analyses of covariance( ANCOVA) of the relationship between the allocation to the male and female reproductive function in A. arbustorum that copulated once, twice or three times (=groups). Thumbnail image of

Discussion

The present study provides to our knowledge for the first time a direct estimate of the reproductive allocation to the male and female function in a simultaneous hermaphrodite land snail with internal fertilization. We found that the reproductive allocation was highly female-biased in A. arbustorum and that an increased mating frequency led to an increased allocation to the male function. The latter finding was predicted by Charnov (1996) and Greeff & Michiels (1999), even though their models considered much higher mating frequencies (5–50 or even an infinite number of copulations). The mating frequency of A. arbustorum in natural populations might not be as high as assumed by these authors. Individuals of A. arbustorum have been observed to mate repeatedly with different partners in the course of a reproductive season in a field cage experiment, but actual data on mating frequency are not available ( Baur, 1988). Mating frequencies have been recorded in two other helicid species. Helix pomatia copulated 2–6 times per year in a Danish population ( Lind, 1988), 2–4 times in a German population ( Tischler, 1973) and Helix aspersa on average three times (maximum seven times) in a British population ( Fearnley, 1993, 1996). Taking into account the short period of activity of snails living in subalpine populations, the assumed range of 1–3 copulations per reproductive season might be reasonable for the animals studied.

The present study showed a strong female skew in sex allocation, which was, however, much larger than predicted by the models of Charnov (1996) and Greeff & Michiels (1999). A possible explanation for this pronounced female skew could be incomplete estimates of reproductive allocation to either function. Moreover, A. arbustorum may not entirely fulfil the assumptions of the models, which consider a higher mating frequency. Actually, there exists no sex allocation model for outcrossing simultaneous hermaphrodites with low mating frequency.

In simultaneous hermaphrodite land snails like A. arbustorum, each mating in the male role caries costs, which includes the costs of courtship behaviour (including the optional dart shooting), spermatophore and sperm production and other possible costs associated with mating. During the long-lasting courtship terrestrial gastropods produce huge amounts of mucus, an energetically expensive behaviour ( Calow, 1977; Davies et al., 1990 ). The present study does not provide any estimate of the actual costs of courtship behaviour. However, our results suggest that these costs may be similar in successive copulations, as courtship and copulation duration remained unchanged with increasing mating experience of the snails. Furthermore, it is not clear whether the costs of courtship can entirely be assigned to the male function. The female function of this outcrossing hermaphrodite needs at least one copulation per reproductive season for the fertilization of the eggs ( Chen & Baur, 1993; Chen, 1994). From the point of view of offspring quality, it has also been suggested that the female function might use (so far unknown) cues from courtship behaviour to assess the quality of the potential mating partner ( Baur et al., 1998 ; Baur, 1998). Indeed, there is some evidence for the occurrence of selective sperm digestion (cryptic female choice; cf. Eberhard, 1996) in A. arbustorum ( Haase & Baur, 1995). If this is the case, the nontrivial costs of courtship behaviour should be distributed to both the male and female function. Finally, in a variety of terrestrial gastropods including A. arbustorum, copulation per se and/or the transfer of the spermatophore stimulates egg production via hormones in both partners ( Saleuddin et al., 1991 ; Baur & Baur, 1992). Thus, a copulation might be beneficial to both functions even if the sperm delivered in repeated copulations are not used by the female function for egg fertilization. It is difficult to assess the different costs associated with mating behaviour and to assign them to a specific sex function in A. arbustorum as well as in any other pulmonate land snail. However, the long-lasting courtship and copulation behaviour suggest that these costs might be substantial. In the present study we measured energy allocation to gamete production rather than the energetic costs of male and female sexual function in the broad sense.

A fundamental assumption of sex allocation theory in simultaneous hermaphrodites is a trade-off between male and female function, i.e. the animal has a fixed amount of resources to allocate between the gender ( Charnov, 1982). A hermaphrodite that suppresses the use of one sex function may automatically free resources for the other sex function. Trade-offs between the male and female function within a population have been documented in some plant species (e.g. Garnier et al., 1993 ; Ashman, 1999), but not in others (e.g. Damgaard & Loeschcke, 1994; Mossop et al., 1994 ). De Visser et al. (1994 ) experimentally suppressed the male function in the simultaneous hermaphrodite freshwater snail Lymnaea stagnalis, which resulted in an increased egg production. This freshwater snail mates frequently and transfers spermatozoa in a sperm suspension. Copulation is unilateral, but reversal of roles occurs in most matings ( van Duivenboden & ter Maat, 1988). On the other hand, Doums & Jarne (1996) found no difference in allocation to the female function between euphallic morphs (regular hermaphrodites) and aphallic morphs (individuals without a male copulatory organ) of Bulinus truncatus.

Resource allocation considerations generate predictions of a negative quantitative relationship between male and female functions. Trade-offs may occur when resources are in short supply. A. arbustorum feeds on vascular plants and dead or senescent herbs, but the snails often supplement their vegetable diet with micro-organisms or carrion ( Frömming, 1954; Grime & Blythe, 1969). A short supply of nitrogen might be critical for the reproduction of this land snail ( White, 1993). Thus, the nitrogen content of spermatophores and eggs might be an appropriate measure for examining a possible trade-off in the allocation to the male and female function. However, we did not find any trade-off between the two functions. In contrast, we found a positive relationship between the number of sperm delivered and the number of eggs produced as well as between resources allocated to the male and female functions. With regard to quality, a ‘good genes’ scenario might be valid: any genes that affect spermatozoa and ova quality in the same direction would lead to a positive association between the two ( Schlichting & Delesalle, 1997). In simultaneous hermaphrodite pulmonates including A. arbustorum spermatozoa and ova are produced simultaneously in the same organ, the so-called ovotestis. It would be most interesting to disentangle possible associations between spermatozoa and ova quality in these snails.

In simultaneous hermaphrodites like A. arbustorum, the optimal allocation to male vs. female function may vary with population density. Snails living in populations with a high density might be exposed to an increased risk of sperm competition ( Birkhead & Møller, 1998; Simmons & Siva-Jothy, 1998). Under these conditions an increased mating frequency is expected and, consequently, a larger allocation to the male reproductive function. Adaptive sex allocation has been shown in simultaneous hermaphrodite chalk bass, Serranus tortugarum ( Petersen & Fischer, 1996) and in the trematode, Echinostoma caproni ( Trouvéet al., 1999 ).

Acknowledgments

We thank A. Baur, A. Erhardt, M. Puurtinen, P. Ward and two anonymous reviewers for discussion and/or constructive comments on the manuscript. Financial support was received from the Swiss National Science Foundation (grant no. 31–53688.98).

Ancillary