Luminous AGB variables in the dwarf Irregular Galaxy, NGC 3109
MMNRAS , 000–000 (0000) Preprint 20 December 2018 Compiled using MNRAS L A TEX style file v3.0
Luminous AGB variables in the dwarf Irregular Galaxy, NGC 3109
John W. Menzies , Patricia A. Whitelock , , Michael W. Feast , and Noriyuki Matsunaga South African Astronomical Observatory, P.O.Box 9, 7935 Observatory, South Africa. Astronomy Department, University of Cape Town, 7701 Rondebosch, South Africa. department of Astronomy, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan.
20 December 2018
ABSTRACT
In a shallow near-infrared survey of the dwarf Irregular galaxy, NGC 3109, near the peripheryof the Local Group, we have found eight Mira variables, seven of which appear to be oxygen-rich (O-Miras). The periods range from about 430 days to almost 1500 days. Because of ourrelatively bright limiting magnitude, only 45 of the more than 400 known carbon stars weremeasured, but none was found to be a large amplitude variable. One of the Miras may be anunrecognised C star. Five of the O-Miras are probably hot-bottom burning stars consideringthat they are brighter than expected from the period–luminosity relation of Miras and that, bycomparison with theoretical evolutionary tracks, they appear to have masses (cid:38) M (cid:12) . A censusof very long period ( P > JHKL photometry is presented for three O-rich long period Miras in theSMC (including a candidate super-AGB star).
Key words: stars: AGB and post-AGB–galaxies: individual: NGC 3109–Local Group–stars:variables: general–infrared: stars
NGC 3109 is a small galaxy, classified as SB(s)m (de Vaucouleurset al. 1991) or dIrr (McConnachie 2012) that is associated with agroup of dwarf irregular galaxies including Sextans A, Sextans B,Antlia and possibly Leo P. NGC 3109 is outside the virial radiusof the Local Group and is probably not bound to it, though it mayhave been near the Milky Way about 7 – 9 Gyr ago (Pawlowski& McGaugh 2014). It was nevertheless included in our infraredsearch at SAAO for long period variables in the Local Group (seee.g. Menzies et al. 2015).Based on the colour of the red giant branch (RGB) (Hidalgoet al. 2008), the metallicity is very low, [Fe / H] = − . ± .
2. Nev-ertheless, a spectroscopic study of blue supergiants and H ii regionsin NGC 3109 (Evans et al. 2007; Hosek et al. 2014; Tramper et al.2014) results in a metallicity that is significantly higher, [Fe / H] = − .
67. Pe˜na et al. (2007) find O / H + = . ± .
07 for H ii regions,in good agreement with that found by Evans et al. (2007), and sug-gest that the current interstellar medium in NGC 3109 is chemicallyvery homogeneous. Thus the young component of NGC 3109 hasan O / H ratio about half that of the SMC.There have been many determinations of distance, which aresummarised by Soszy´nski et al. (2006). We use the modulus, (m − M) = .
6, in this paper. The reddening is small, E B − V ∼ . i envelope (Jobin & Carignan 1990; Carignan et al. 2013). The H i is warped in the outerdisk towards the neighbouring Antlia galaxy to the south-west sug-gesting an interaction in the past, though not at present (Carignanet al. 2013); the disk is seen almost edge-on with an inclination of75 ◦ . The rotation curve is consistent with a dark-matter halo thatdominates at all radii. Demers et al. (2003) found more than 400C stars in the galaxy, mostly confined to the plane of the disk, andthere does not appear to be an extensive halo. They find a C / M ratioof 1.75, typical for a metal-poor galaxy such as this (see e.g. Cioni2009).The present investigation extends our study of AGB variablesin the Local Group, which to date has covered dwarf spheroidals(Leo I: (Menzies et al. 2010), Fornax: (Whitelock et al. 2009),Sculptor: (Menzies et al. 2011)) and dwarf irregulars (NGC 6822:(Whitelock et al. 2013), IC 1613: (Menzies et al. 2015), Sgr DIG:(Whitelock et al. 2018)). Our project is to investigate AGB vari-ables in di ff erent environments with two primary aims: first, to testmodels of AGB evolution (e.g. Marigo et al. 2017), and secondlyto investigate how well Miras can be used as fundamental distanceindicators. Short period Miras, which occur in Galactic globularclusters (e.g. Feast et al. 2002) are reasonably well understood andare proving to be useful distance indicators (e.g. Whitelock et al.2008a; Whitelock & Feast 2014; Huang et al. 2018; Yuan et al.2018), that will be particularly important in the JWST era, becausethey are bright infrared sources.Miras with periods significantly over 400 days, which havemore massive progenitors, remain very poorly understood. Theseare often strong mid-infrared sources (Riebel et al. 2015; Whitelock c (cid:13) a r X i v : . [ a s t r o - ph . S R ] D ec Menzies et al. . − M (cid:12) (Karakas et al. 2017), that de-termines which stars become supernovae and which end their livesas white dwarfs. The input physics is uncertain, so models di ff ersignificantly from one to another. Observations have yet to isolatea definite example, but the best candidates are the very luminous,large amplitude, long-period variables of the type we investigatehere.Given the distance to NGC 3109, we are only sensitive to thebrightest variables, so our objective in the present study is to inves-tigate the most massive of the AGB variables. A deeper survey isrequired to get a complete sample. Our observations were made with the SIRIUS camera on theJapanese-SAAO IRSF telescope at Sutherland. The camera pro-duces simultaneous J , H and K S images covering a 7.2 × / pixel. NGC 3109 is extendedin approximately the east-west direction and it was necessary to use3 pointings to cover the bulk of the galaxy. These were centred at α (2000.0) = δ (2000.0) = –26:09:31.9, (field 1), and dis-placed by ± . K S band, and typically 50 or30 exposures, respectively, were combined to produce the final im-ages. Photometry was performed using dophot in ‘fixed-position’mode, using the best-seeing H -band image as a template. Aladin(Bonnarel et al. 2000) was used to correct the WCS on each tem-plate and RA and Dec were determined for each measured star.This allowed a cross-correlation to be made with the 2MASS cata-logue, and photometric zero points were determined by comparisonof our photometry with that of 2MASS. With at least 24, mostlyforeground, stars in the range K S = . − . ± .
01 mag.
Fig. 1 shows the K S − ( J − K S ) and ( J − H ) − ( H − K S ) diagrams.The variables and C stars are highlighted. In the two-colour dia-gram, the box outline, or ‘Mira co ffi n’ (Feast et al. 1984), showsthe region where O-rich Miras in the Galaxy are found (the orig-inal coordinates of the box edges were on the SAAO system andhave been transformed to the 2MASS system for this diagram us-ing relations in Carpenter (2001)). The box is displaced with re-spect to the locus of normal giants in both J − H and H − K S dueto the dominance of H O absorption in the near infrared. Galacticcarbon-rich stars (including Miras), with little or no water absorp-tion, fall along the upper left edge of the box, and extend it to larger J − H and H − K S when they experience significant circumstellar reddening (Feast et al. 1982). Six of the newly-discovered Mirasfall in the box.Of the more than 400 C stars found in the galaxy by Demerset al. (2003) only 45 appear in our photometric catalogue becauseof our relatively bright limiting magnitude. A colour-magnitude di-agram published by G´orski et al. (2011) shows that the bulk of theC stars must be about one magnitude fainter than our observinglimit. From the distribution of stars in the two diagrams, it seems clearthat there are many field stars present. It is instructive to comparethese diagrams with what is expected in the field based on a trile - gal (Girardi et al. 2005) simulation for the same area as our pho-tometry covers. This is shown in Fig. 2, where the trilegal pointsare shown in magenta superposed on our diagrams. It should benoted that no observational errors have been applied to the trile - gal data.It appears that, in our sample of the NGC 3109 field, unam-biguous members lie to the red of J − K S ∼ .
9. Thus, as expected,our survey is very shallow and comprises mostly the late-type gi-ants on the AGB and supergiants (if present). The same conclusioncould be reached by making J − H and H − K S cuts in the two-colourdiagram to isolate the obvious field dwarf and giant sequences. As stated earlier, the purpose of our study was to search for andcharacterise long period variables in NGC 3109. We have discov-ered nine large amplitude variables – eight are apparently Miras,while one is unusual, being blue and having a peculiar light curveas discussed in section 4.2.
Median standard deviations were determined at half-magnitude in-tervals for each of the wavebands represented in Fig. 3. Lines weredrawn at twice these median values (see the figure) and a star wasconsidered variable if for all of J , H and K S its standard deviationlay above the relevant line. We used Lomb-Scargle periodogramsto find possible periods. Because the data points are not well dis-tributed in time and for some objects are relatively noisy, we foundthat the best results were obtained by adopting the multiband pe-riodogram approach of VanderPlas & Ivezi´c (2015) whereby all 3wavebands are fitted simultaneously. The stars for which we foundperiods are shown as asterisks in Fig. 3, where we plot standarddeviation against magnitude for our catalogue. The stars with sim-ilarly large standard deviations may be variable but are either notperiodic or may actually be marginally resolved background galax-ies that appear to vary because of seeing fluctuations from image toimage.The light curves of the eight periodic variables of the eightperiodic variables are shown in Fig. 4 with single-period sinusoidsoverplotted on the observations. Mean magnitudes estimated fromthe lightcurve fits are listed, together with the derived periods andpeak-to-valley amplitudes ( ∆ J , ∆ H , ∆ K S ), in Table 1. MNRAS000
Median standard deviations were determined at half-magnitude in-tervals for each of the wavebands represented in Fig. 3. Lines weredrawn at twice these median values (see the figure) and a star wasconsidered variable if for all of J , H and K S its standard deviationlay above the relevant line. We used Lomb-Scargle periodogramsto find possible periods. Because the data points are not well dis-tributed in time and for some objects are relatively noisy, we foundthat the best results were obtained by adopting the multiband pe-riodogram approach of VanderPlas & Ivezi´c (2015) whereby all 3wavebands are fitted simultaneously. The stars for which we foundperiods are shown as asterisks in Fig. 3, where we plot standarddeviation against magnitude for our catalogue. The stars with sim-ilarly large standard deviations may be variable but are either notperiodic or may actually be marginally resolved background galax-ies that appear to vary because of seeing fluctuations from image toimage.The light curves of the eight periodic variables of the eightperiodic variables are shown in Fig. 4 with single-period sinusoidsoverplotted on the observations. Mean magnitudes estimated fromthe lightcurve fits are listed, together with the derived periods andpeak-to-valley amplitudes ( ∆ J , ∆ H , ∆ K S ), in Table 1. MNRAS000 , 000–000 (0000) uminous AGB variables in NGC 3109 s K s s J − H Figure 1. (Left) Colour-magnitude diagram for NGC 3109. The small dots (grey) show the field and galaxy non-variable stars. Triangles (red) show knownor suspected C stars, star symbols (green) are the variables listed in Table 1. On the basis of their positions in the two-colour diagram the points marked asinverted triangles (cyan) are probably background galaxies. (Right) Two-colour diagram for NGC 3109. Coloured symbols are as in the colour-magnitudediagram. The continuous and dashed lines show the loci of field dwarfs and giants, respectively, from Bessell & Brett (1988) transformed from the Glasssystem to the 2MASS system. The region occupied by O-rich Miras in the Galaxy is shown by the closed box (Feast et al. 1984), the so-called ‘Mira co ffi n’. Table 1.
JHK S Photometry of Nine Large Amplitude VariablesName RA Dec Period ∆ J ∆ H ∆ K S K S J − H H − K S J − K S BC m bol note(2000.0) day ––––––––––––––––––––––––––––––mag––––––––––––––––––––––––––––––1067 150.73837 -26.14197 591 0.88 1.22 1.00 16.102 0.682 0.321 1.003 3.20 19.301112 150.80582 -26.15483 583 0.49 0.59 0.62 16.039 0.782 0.355 1.137 3.20 19.241153 150.75345 -26.17370 678 0.91 0.97 0.93 16.284 0.685 0.398 1.083 3.20 19.481224 150.82961 -26.14808 680 0.73 0.85 0.74 16.269 0.657 0.456 1.113 3.20 19.472075 150.91414 -26.17125 1109 1.69 1.34 1.34 16.556 1.169 0.440 1.609 3.19 19.742081 150.86487 -26.15757 434 0.71 0.74 0.60 16.766 0.828 0.357 1.185 3.20 19.973064 150.65652 -26.14415 562 0.37 0.42 0.47 15.839 0.874 0.273 1.147 3.20 19.043089 150.59729 -26.17039 1486 1.24 1.18 0.98 15.921 0.939 0.582 1.521 3.19 19.11 a ba period about the same as the length of the data train. b Period and amplitudes after removal of linear trend. See text in section 4.2.
In the case of object J , H , K S magnitudes changedby − . , − .
55 and − .
49, respectively, over the period coveredby our data. After linear slopes were removed, the period-findingprogram found a period of 276 day would fit the data, though the K S data are consistent with no variation. The data for this ob-ject are shown in Fig. 5, together with the fitted two-componentlightcurves.This variable is clearly blue on the basis of our photometry ( J − K S = . , K S = . J − K S = . , K S = . MNRAS , 000–000 (0000)
Menzies et al. s K s s J − H Figure 2.
Colour-magnitude and two-colour diagrams for NGC 3109 with stars from a trilegal model superimposed as squares (magenta). Symbols andcolours otherwise as in Fig. 1. The region between the dashed lines is where O-rich stars on the AGB with [Fe / H] = − . where it has clearly blue colours at epoch JD2456096.5, but isflagged as extended.Our measurement appears to refer to the combined light of thered supergiant and another object. The 276-day variation might bedue to the supergiant, but assuming that to have not varied since theepoch of the 2MASS measurement, the contribution of the secondobject at our mid observational epoch would have been ( J − H = − . H − K S = − . J − K S = − . K S = .
43) but the2MASS errors are too large to attach any real significance to thisresult, save to say that the object is bluer than the supergiant. Takenat face value these colours suggest a significant emission line fluxfrom the blue object.In the ACS Nearby Galaxy Survey Treasury images ofNGC 3109 (Holtzman et al. 2006), there is an object about 1.15arcsec north-east of the supergiant; it is object F W − F W = .
1, correspondingto V − I = . Intrinsic carbon stars form following third dredge-up (3DUP) whensu ffi cient carbon reaches the stellar surface to change the surfacechemistry from C / O < / O > O , TiO etc to CN, CH, C etc. (notethat CO is always present and uses up most of either C (in O-richstars) or O (in C-rich) stars). The amount of 3DUP depends on themass and metallicity of the star, and the amount of 3DUP requiredto change the surface chemistry depends on the initial oxygen abun-dance, as the more oxygen initially present the more carbon is re-quired to move the ratio to C / O >
1. The relative numbers of C-and O-rich AGB stars is theoretically a sensitive probe of mass andmetallicity (see, e.g. Marigo et al. 2017, fig. 10), although it hasyet to be calibrated against observations over large ranges of thesequantities.The relative numbers of C stars and M stars is often usedas a proxy for metallicity and methods such as that devised byCioni et al. (2006) are used to establish the region in a colour-magnitude diagram (e.g. Jones et al. 2018) where the O- and C-rich stars should be expected. This is illustrated in Fig. 2 wherethe area in which O-rich stars are anticipated at [Fe / H] = − . ± / H] moves the lines by ± .
06 in J − K S .O-rich stars should fall between these lines, while C stars shouldfall to the right. Note that if this method was used on NGC 3109then almost one third of the C-stars we observed would have beenclassified as O-rich. Thus this is not an e ff ective way of isolating Cstars in galaxies that are significantly di ff erent from the MagellanicClouds, where the method was calibrated. In contrast, the Marigo MNRAS000
06 in J − K S .O-rich stars should fall between these lines, while C stars shouldfall to the right. Note that if this method was used on NGC 3109then almost one third of the C-stars we observed would have beenclassified as O-rich. Thus this is not an e ff ective way of isolating Cstars in galaxies that are significantly di ff erent from the MagellanicClouds, where the method was calibrated. In contrast, the Marigo MNRAS000 , 000–000 (0000) uminous AGB variables in NGC 3109
12 13 14 15 16 17 18 19K S s t d . d e v ( m a g ) Figure 3.
Standard deviation versus magnitude for the catalogue. Periodicvariables and star models discussed in section 6 fit our observations of C stars ratherwell.None of our variables matches any star in the C star list ofDemers et al. (2003). None of the 45 C stars in this list that wehave measured is a large amplitude variable. Our limiting magni-tude is rather bright, being set by the faintest measurable stars onthe reference H image, so we have missed most of the C stars inour field. Close inspection of the K S images showed that a further172 C stars were visible on at least 10 frames. We used Sextractor(Bertin & Arnouts 1996) to measure these stars on as many framesas possible. The precision of the photometry is low ( ± . It is instructive to compare the variables in NGC 3109 with those inother dwarf irregular galaxies, namely, NGC 6822, WLM, IC 1613and the SMC. In Fig. 6 we have added our photometry for the redsupergiants from Levesque & Massey (2012) and for the O-Mirasfrom Whitelock et al. (2013) in NGC 6822, our unpublished datafor the red supergiants (Levesque & Massey 2012) and newly dis-covered variables in WLM (Menzies, in preparation), as well as forthe four O-rich Miras in IC 1613 (Menzies et al. 2015). We havecorrected the NGC 6822 data for the di ff erential distance modulusof 2.03 mag and an assumed reddening for that galaxy of E( J − K S ) = K S magnitudes have been adjusted downward by 0.7 and1.42 mag, respectively. Data for the SMC variables are taken fromthe discussion in section 6. The five variables in NGC 3109 withperiods between 430 and 680 days lie to the red of the supergiantsin the colour-magnitude and two-colour diagrams. They occupy thesame regions as do the HBB Miras in NGC 68222 and IC 1613. The log period-apparent K S luminosity diagram based on the datafrom Table 1 is compared with the PL relation for the LMC (White-lock et al. 2008b) shifted from a distance modulus of 18.5 to ourassumed value of 25.6 for NGC 3109 is shown in Fig. 7. The lin-ear relation was established by Whitelock et al. (2008b) for log P (cid:54) = > Star formation in NGC 3109 has taken place in two episodes (Weiszet al. 2011). In the first, almost 80% of the galaxy’s stars wereformed by about 10 Gyr ago; these presumably are the giants andfainter stars with a very low metallicity. In more recent times, sinceabout 2 Gyr ago, stars have been forming at a slow rate with pre-sumably a steady increase in metallicity to the level seen in theblue supergiants (e.g. Tramper et al. 2014). Thus we would expectthere to be a range of metallicity and age amongst the AGB stars.To illustrate this, we have obtained a selection of the latest parsec +colibri tracks (Marigo et al. 2017) and overlaid them on the colour-magnitude diagram of G´orski et al. (2011) in Fig. 8. This can onlybe illustrative, but suggests a plausible range of ages and metal-licities for the AGB of NGC 3109. Note that the 0.398 Gyr trackcovers the C stars that we measured, while the 0.798 Gyr trackoverlies the bulk of the fainter, red AGB stars covered by Gorsk´ı’smeasurements.To investigate the nature of the variables further we have useda small set of evolutionary tracks for the thermally-pulsing phaseof the AGB (Marigo et al. 2017) (kindly supplied by Paola Marigo)covering the mass range from 2.6 M (cid:12) to 5 M (cid:12) , for metallicitiesZ = (0.0005, 0.001, 0.004). Five of the six proposed HBB variablesare well represented by a track for a star with an O-rich atmosphere,M = (cid:12) and Z = (cid:12) , Z = ∼ (cid:12) star on a tracksoon after becoming a C star. In the two-colour diagram, Fig 1, itis outside the O-Mira box, and at the blue end of the C-star region,though it is not identified as a C star by Demers et al. (2003). Itmay be in a transitional evolutionary phase. The longest period star( (cid:12) star with Z in the MNRAS , 000–000 (0000)
Menzies et al. (a) 2081 (434 d) (b) 3064 (562 d)(c) 1112 (583 d) (d) 1067 (591 d)(e) 1153 (678 d) (f) 1224 (680 d)(g) 2075 (1109 d) (h) 3089 (1486 d)
Figure 4.
JHK S light curves for eight long period variables. Periods are shown in brackets following the star number. Curves are single-period sinusoidal fitsto the data. MNRAS000
JHK S light curves for eight long period variables. Periods are shown in brackets following the star number. Curves are single-period sinusoidal fitsto the data. MNRAS000 , 000–000 (0000) uminous AGB variables in NGC 3109 K H J Figure 5.
JHK S Lightcurves superimposed on the data for range 0.0005–0.004. Its observed period far exceeds the maximumpredicted for any of these tracks.In the case of star ∼ (cid:12) andZ = = (cid:12) , and metallicity Z = = ff erent groups on the basisof a combination of Gaia and near-infrared photometry, which wehave applied to variables in NGC 3109 and in the LMC, the SMCand some of the Local Group galaxies that we have observed in thepast. For NGC 3109, we find general agreement between the clas-sification based on this approach and that based on the evolutionarytracks.
There are only a few large amplitude AGB variables known withprimary pulsation periods over 1000 days, partly because they arerare, but the limited patience of observers and TACs also makestheir characterization di ffi cult. These stars are of special interestbecause they are uncommon and will include short-lived evolution-ary phases of the most massive stars that experience the AGB. Inour paper on Sgr dIG (Whitelock et al. 2018) we discussed a Mirawith a period of P =
950 days and suggested it was in a post-HBB phase and leaving the AGB. In view of the fact that we havefound two even longer period stars in NGC3109, it is appropriateto briefly review how many very long period stars there are in theLocal Group and what we know about their evolution. We start witha brief census of P > LMC:
Wood et al. (1992) measured periods of over 1000 daysfor eight IRAS sources in the LMC and detected OH emissionin four of them, confirming their O-rich nature. Whitelock et al.
Table 2.
Very Long Period Variables in the LMCName RA Dec Period(2000.0) (day)Carbon richIRAS 05125–7035 078.003208 -70.540000 1115IRAS 05278–6942 081.850458 -69.662472 1001IRAS 05506–7053 087.485500 -70.886583 1026IRAS 05509–6956 087.608770 -69.934250 1052IRAS 05568–6753 089.161500 -67.892889 1209Oxygen richIRAS 04407–7000 070.118667 -69.920417 1148IRAS 04498–6842 072.422792 -68.630944 1256IRAS 04509–6922 072.668583 -69.292194 1271IRAS 04516–6902 072.870792 -68.963889 1095IRAS 04545–7000 073.541875 -69.932833 1254MSX LMC 1210 073.835417 -68.377583 1050HV 888 076.058875 -67.270639 1005MSX LMC 642 082.200708 -71.041361 1133IRAS 05298–6957 082.352600 -69.920468 1265MSX LMC 807 083.154833 -67.115667 1069IRAS 05329–6708 083.213875 -67.114444 1303IRAS 05402–6956 084.937000 -69.921667 1362IRAS 05558–7000 088.837625 -70.000833 1175 (2003) confirmed the long periods for six of these and measured an-other three with P > . On the basis of their colours Soszy´nskiet al. (2009) suggested that seven of the variables with long pe-riods were O-rich and one, IRAS05402–6956 (P = SMC:
The OGLE group (Soszy´nski et al. 2011) measured pe-riods for three O-rich Miras in the SMC of between 1062 and 1859days. All three Miras have confirmed M-type spectra (Elias et al.1980; Groenewegen et al. 1998; van Loon et al. 2008).
JHKL pho-tometry for these three stars is given in Appendix B and a summaryof their mean colours and amplitudes is given in Table 3, where theperiods are taken from OGLE (Soszy´nski et al. 2011), but agreewith what is derived from the measurements in the Appendix.HV 11417 is sometimes classed as a supergiant, but thesecolours put it in the region of the two-colour diagram occupied byMiras (Feast et al. 1984) which, together with its large amplitude,supports the AGB classification favoured recently (Soszy´nski et al.2011; Groenewegen & Sloan 2018). The other two stars have thecolours of Miras reddened by circumstellar reddening and couldbe either O- or C-rich. All three stars are shown in Fig. A1 whereIRAS00483–7347 falls in the region dominated by C stars, showingthat the classification scheme based on
Gaia and near infrared pho-tometry breaks down (as do other photometrically-based schemes)for O-rich stars with high mass-loss rates. Transforming the pho-tometry to the 2MASS system following Carpenter (2001) gives The OGLE catalogue gives multiple periods for many stars. Here we onlyconsider the dominant one. updated: http: // / ∼ jmc / / v3 / transformations / MNRAS , 000–000 (0000)
Menzies et al. s K s s J − H Figure 6.
Colour-magnitude (left) and two-colour (right) diagrams for NGC 3109 with points for NGC 6822, WLM, IC 1613 and the SMC added. Allmagnitudes have been shifted to the distance modulus of NGC 3109. The X symbols (orange) show the positions of the M-type supergiants from Levesque& Massey (2012) and the diamond symbols (orange) the O-Miras from Whitelock et al. (2013), in the galaxy NGC 6822 adjusted for reddening. The plussymbols and squares (magenta) show the supergiants and variables, respectively, for WLM. The filled circles (cyan) show the IC 1613 variables. The threeSMC variables are shown as thin diamonds (black). Symbols and colours otherwise as in Fig. 1.
Table 3.
SAAO
JHKL
Photometry of SMC Variables (all O-rich)Name RA Dec Period ∆ J ∆ H ∆ K K J − H H − K J − K K S * J − K S * (2000.0) day magHV11417 015.20071 -72.85056 1092 1.26 1.48 1.50 8.86 0.97 0.47 1.44 8.86 1.35GM103 012.62764 -72.85830 1062 1.27 1.34 1.18 8.91 1.10 0.61 1.70 8.91 1.60IRAS00483–7347 012.52963 -73.52377 1859 0.98 0.79 0.57 8.71 1.72 1.11 2.83 8.73 2.67 * Transformed from the SAAO to the 2MASS system. the values listed in the last two columns of the table. These areshown in Fig. A1 for comparison with the NGC 3109 Miras, as-suming a distance modulus of 18.9 for the SMC.Groenewegen & Sloan (2018) suggest that IRAS 00483–7347(MSX SMC 055) is the best Magellanic Cloud candidate for asuper-AGB star, based on its high luminosity ( M bol = − . M = . ± . M (cid:12) and an initial mass around 1 M (cid:12) larger than this.A high mass is independently supported by the measured high ru-bidium abundance (Garc´ıa-Hern´andez et al. 2009). The Galaxy:
The GCVS (Samus’ et al. 2017) lists 11 Mi-ras with periods over 1000 (up to 1994) days. Only one of these,V829 Cas, with a period of 1060 days (Groenewegen et al. 1998),is a C star (Zuckerman & Dyck 1986). This is the longest periodknown for a C-rich Mira in the Galaxy, and the only one above1000 days. The models discussed by Marigo et al. (2017) do notproduce C stars at high metallicities, so we would anticipate that such stars are less common in the Galaxy, particularly among thehigher-mass longer-period AGB stars, than e.g., in the MagellanicClouds.Four of the other ten GCVS variables are either not Miras orhave periods much shorter than 1000 days. The ASAS online cata-logue (Pojmanski 1997) convincingly provides shorter periods forthree of them, CD Pup, V1156 Sgr and V581 CrA (the last is ac-tually an RCB star (Miller et al. 2012)) and the AAVSO catalogue(Watson et al. 2006) gives a short period for EI Sct. The other sixlong period Miras in the GCVS are OH / IR stars, so they are defi-nitely AGB stars and their long periods are at least plausible; theseare listed in Table 4 with their periods and the uncertainty on theperiod ( ∆ P ) where it is given in the reference. The GCVS also liststhree possible Miras with P > / IR star and is included in the Table. Very little is
MNRAS000
MNRAS000 , 000–000 (0000) uminous AGB variables in NGC 3109 K S AA O Figure 7.
Period-luminosity diagram for NGC 3109 Mira variables (green ∗ symbols). The solid line shows the relation for the LMC Whitelock et al.(2008b) with an assumed distance modulus of 25.6 for NGC 3109while thedashed line shows the PL relation of Ita & Matsunaga (2011) for stars withlog P > Figure 8.
Comparison of isochrones with the observed colour-magnitudediagram. The background points are from G´orski et al. (2011), the ∗ sym-bols (green) are our variables, and the (red) triangles show the C starswe measured in the present study. The near-vertical tracks are isochronesfor, from left to right, ages of 0.398 (blue), 0.798 (magenta) and 10.0 Gyr(green), respectively. The isochrones have di ff erent metallicities, decreas-ing from left to right, of Z = known about the third, CP Sco, but the AAVSO list a short period,so it is not included here.There are another nine OH / IR sources for which periods over1000 days have been determined, listed in Table 4, and several ofthem already have GCVS designations. So there are at least 16 O-rich Miras with periods over 1000 days in the Galaxy.In addition to the variables discussed above there are sev-
Figure 9.
Evolutionary track (left panel) and period evolution (right panel)for a 5 M (cid:12) star with Z = ∗ . eral other OH-variables within the Galaxy with periods over 1000days (a few over 2000 days) (e.g., Herman & Habing 1985; vanLangevelde et al. 1990), most of which are probably massive AGBstars (some may be supergiants), but have no optical or near-infrared counterpart, presumably because of circumstellar and / orinterstellar absorption (they are close to the Galactic plane). So,in summary, within the Galaxy there is one C star and at least 16O-rich stars with periods over 1000 days. The census is obviouslyincomplete, but the total number cannot be very large.Long periods among Miras are rare, either because they are ina very short-lived evolutionary phase (see, e.g. Marigo et al. 2017)or because they only occur among the most massive, and there-fore the rarest, of the AGB stars, or both. As summarized by Feast(2009), we understand that the initial mass of Miras with periodsless than about 400 days is a function of their pulsation period. Thesituation is less clear for longer period stars, some of which experi-ence HBB (see, e.g. Whitelock 2017; Whitelock et al. 2018). Verylong period OH-variables, which can be detected to very large dis-tances, are found only close to the Galactic plane, supporting theview that they have progenitors that are significantly more massivethan shorter period Miras.It is therefore very interesting that we find two very long pe-riod stars in NGC 3109. If J − K S ) than Miras of comparable period inthe Galaxy or the LMC, indicating lower circumstellar extinctionand weaker stellar winds. Whitelock et al. (2018) discussed a longperiod ( ∼
950 day) O-rich Mira in Sgr dIG, emphasizing how un-usual it was to find such a long period Mira with such blue colours.Using stellar evolution models they suggested that its progenitorhad a mass around 5 M (cid:12) , and that it was close to the end of itsAGB evolution. It is interesting that we find a very similar star inNGC 3109, which galaxy has about 20 times the mass in the formof stars than does Sgr dIG (McConnachie 2012) (there is insu ffi -cient information to comment on the relative number of stars ofthe appropriate age to produce long period Miras). It is not entirelyclear what metallicity we should associate with these, presumably MNRAS , 000–000 (0000) Menzies et al.
Table 4.
Long period Miras in the Galaxyname P ∆ P Ref otherV0829 Cas* 1060 1 IRAS01144 + + + + + + + + + + + massive, AGB stars, but probably not as low as in Sgr dIG. Themetallicity is important as it a ff ects the mass-loss rates of O-richstars much more strongly than those of C-rich ones (Wood et al.1992; Matsuura et al. 2005). As mentioned above, there is strongevidence that the young stars in NGC 3109 have higher metallici-ties than do those on the giant branch.This is important in the context of the conclusions of Goldmanet al. (2018) who suggest that the dusty winds, which are generallyfound towards the end of the evolution of the most massive coolstars, are curtailed at low metallicity (see also Wood et al. 1992;Matsuura et al. 2005, and references therein). It is also potentiallyimportant in the context of electron-capture supernova (ECSNe)(Langer 2012, section 7.1 and references therein), the progenitorsof which are probably massive AGB stars. The frequency of ECSNemay depend on metallicity, if at low metallicity the stellar winds areweaker, so that evolution is not terminated by mass loss and the corehas time to grow large enough for the star to become a supernova.The discovery of two luminous long period AGB variables in themetal weak environment of NGC 3109 lends support to this channelfor ECSNe. ACKNOWLEDGEMENTS
JWM and PAW wish to thank Prof Rolf Kudritski for providingaccess to the facilities of MIAPP, where part of the work on thispaper was done. This research has made use of Aladin (Bonnarelet al. 2000). This material is based upon work supported finan-cially by the South African National Research Foundation. Someof the data presented in this paper were obtained from the Mikul-ski Archive for Space Telescopes (MAST). STScI is operated bythe Association of Universities for Research in Astronomy, Inc.,under NASA contract NAS5-26555. This publication makes use ofdata products from the Two Micron All Sky Survey, which is a jointproject of the University of Massachusetts and the Infrared Process-ing and Analysis Center / California Institute of Technology, fundedby the National Aeronautics and Space Administration and the Na-tional Science Foundation. The IRSF project is a collaboration be- tween Nagoya University and the SAAO supported by the Grants-in-Aid for Scientific Research on Priority Areas (A) (no. 10147207and no. 10147214) and Optical & Near-Infrared Astronomy Inter-University Cooperation Program, from the Ministry of Education,Culture, Sports, Science and Technology (MEXT) of Japan andthe National Research Foundation (NRF) of South Africa. Thisresearch has made use of the VizieR catalogue access tool, CDS,Strasbourg, France. The original description of the VizieR servicewas published in A&AS 143, 23. This work has made use of datafrom the European Space Agency (ESA) mission
Gaia ( ), processed by the Gaia
Data Pro-cessing and Analysis Consortium (DPAC, ). Funding for the DPAChas been provided by national institutions, in particular the insti-tutions participating in the
Gaia
Multilateral Agreement. We areparticularly grateful to Paola Marigo who generated evolutionarytracks for us. We thank Serge Demers for promptly providing a cat-alogue of C stars in the region of variable araucaria near-infrared data. Fortheir contributions to the original IRSF observations, we wish tothank Toshihiko Tanab´e, Yoshifusa Ita, Shogo Nishiyama, YasuakiHaba and Enrico Olivier. The following people contributed obser-vations to the data in the Appendix: Enrico Olivier, Jacco van Loon,Albert Zijlstra.
REFERENCES
Bertin E., Arnouts S., 1996, A&AS, 117, 393Bessell M. S., Brett J. M., 1988, PASP, 100, 1134Bonnarel F., et al., 2000, A&AS, 143, 33Boyer M. L., et al., 2017, ApJ, 851, 152Carignan C., Frank B. S., Hess K. M., Lucero D. M., RandriamampandryT. H., Goedhart S., Passmoor S. S., 2013, AJ, 146, 48Carpenter J. M., 2001, AJ, 121, 2851Carter B. S., 1990, MNRAS, 242, 1Chambers K. C., et al., 2016, preprint, ( arXiv:1612.05560 )Cioni M.-R. L., 2009, A&A, 506, 1137Cioni M.-R. L., Girardi L., Marigo P., Habing H. J., 2006, A&A, 448, 77Davidge T. J., 2018, ApJ, 856, 129Demers S., Battinelli P., Letarte B., 2003, A&A, 410, 795Elias J. H., Frogel J. A., Humphreys R. M., 1980, ApJ, 242, L13Evans C. J., Bresolin F., Urbaneja M. A., Pietrzy´nski G., Gieren W., Ku-dritzki R.-P., 2007, ApJ, 659, 1198Feast M. W., 2009, in AGB Stars and Related Phenomena. p. 48( arXiv:0812.0250 )Feast M. W., Robertson B. S. C., Catchpole R. M., Lloyd Evans T., GlassI. S., Carter B. S., 1982, MNRAS, 201, 439Feast M. W., Whitelock P. A., Catchpole R. M., Carter B. S., 1984, TheObservatory, 104, 217Feast M., Whitelock P., Menzies J., 2002, MNRAS, 329, L7Gaia Collaboration et al., 2016, A&A, 595, A1Gaia Collaboration et al., 2018, A&A, 616, A1Garc´ıa-Hern´andez D. A., et al., 2009, ApJ, 705, L31Girardi L., Groenewegen M. A. T., Hatziminaoglou E., da Costa L., 2005,A&A, 436, 895Goldman S. R., et al., 2017, MNRAS, 465, 403Goldman S. R., et al., 2018, MNRAS, 473, 3835G´orski M., Pietrzy´nski G., Gieren W., 2011, AJ, 141, 194Groenewegen M. A. T., Sloan G. C., 2018, A&A, 609, A114Groenewegen M. A. T., Whitelock P. A., Smith C. H., Kerschbaum F., 1998,MNRAS, 293, 18Herman J., Habing H. J., 1985, Astronomy and Astrophysics SupplementSeries, 59, 523Hidalgo S. L., Aparicio A., Gallart C., 2008, AJ, 136, 2332MNRAS000
Bertin E., Arnouts S., 1996, A&AS, 117, 393Bessell M. S., Brett J. M., 1988, PASP, 100, 1134Bonnarel F., et al., 2000, A&AS, 143, 33Boyer M. L., et al., 2017, ApJ, 851, 152Carignan C., Frank B. S., Hess K. M., Lucero D. M., RandriamampandryT. H., Goedhart S., Passmoor S. S., 2013, AJ, 146, 48Carpenter J. M., 2001, AJ, 121, 2851Carter B. S., 1990, MNRAS, 242, 1Chambers K. C., et al., 2016, preprint, ( arXiv:1612.05560 )Cioni M.-R. L., 2009, A&A, 506, 1137Cioni M.-R. L., Girardi L., Marigo P., Habing H. J., 2006, A&A, 448, 77Davidge T. J., 2018, ApJ, 856, 129Demers S., Battinelli P., Letarte B., 2003, A&A, 410, 795Elias J. H., Frogel J. A., Humphreys R. M., 1980, ApJ, 242, L13Evans C. J., Bresolin F., Urbaneja M. A., Pietrzy´nski G., Gieren W., Ku-dritzki R.-P., 2007, ApJ, 659, 1198Feast M. W., 2009, in AGB Stars and Related Phenomena. p. 48( arXiv:0812.0250 )Feast M. W., Robertson B. S. C., Catchpole R. M., Lloyd Evans T., GlassI. S., Carter B. S., 1982, MNRAS, 201, 439Feast M. W., Whitelock P. A., Catchpole R. M., Carter B. S., 1984, TheObservatory, 104, 217Feast M., Whitelock P., Menzies J., 2002, MNRAS, 329, L7Gaia Collaboration et al., 2016, A&A, 595, A1Gaia Collaboration et al., 2018, A&A, 616, A1Garc´ıa-Hern´andez D. A., et al., 2009, ApJ, 705, L31Girardi L., Groenewegen M. A. T., Hatziminaoglou E., da Costa L., 2005,A&A, 436, 895Goldman S. R., et al., 2017, MNRAS, 465, 403Goldman S. R., et al., 2018, MNRAS, 473, 3835G´orski M., Pietrzy´nski G., Gieren W., 2011, AJ, 141, 194Groenewegen M. A. T., Sloan G. C., 2018, A&A, 609, A114Groenewegen M. A. T., Whitelock P. A., Smith C. H., Kerschbaum F., 1998,MNRAS, 293, 18Herman J., Habing H. J., 1985, Astronomy and Astrophysics SupplementSeries, 59, 523Hidalgo S. L., Aparicio A., Gallart C., 2008, AJ, 136, 2332MNRAS000 , 000–000 (0000) uminous AGB variables in NGC 3109 Holtzman J. A., Afonso C., Dolphin A., 2006, ApJS, 166, 534Hosek Jr. M. W., et al., 2014, ApJ, 785, 151Huang C. D., et al., 2018, ApJ, 857, 67Ita Y., Matsunaga N., 2011, MNRAS, 412, 2345Jim´enez-Esteban F. M., Garc´ıa-Lario P., Engels D., Manchado A., 2006,A&A, 458, 533Jobin M., Carignan C., 1990, AJ, 100, 648Jones O. C., Maclay M. T., Boyer M. L., Meixner M., McDonald I.,Meskhidze H., 2018, ApJ, 854, 117Karakas A., Ventura P., Dell’Agli F., Di Criscienzo M., 2017, Mem. Soc.Astron. Italiana, 88, 226Kroll P., Samus N., Volkov I., 2003, Information Bulletin on Variable Stars,5441, 1Langer N., 2012, Annual Review of Astronomy and Astrophysics, 50, 107Lebzelter T., Mowlavi N., Marigo P., Pastorelli G., Trabucchi M., WoodP. R., Lecoeur-Ta¨ıbi I., 2018, A&A, 616, L13Levesque E. M., Massey P., 2012, AJ, 144, 2Marigo P., et al., 2017, ApJ, 835, 77Matsuura M., et al., 2005, A&A, 434, 691McConnachie A. W., 2012, AJ, 144, 4Menzies J. W., Whitelock P. A., Feast M. W., Matsunaga N., 2010, MN-RAS, 406, 86Menzies J. W., Feast M. W., Whitelock P. A., Matsunaga N., 2011, MN-RAS, 414, 3492Menzies J. W., Whitelock P. A., Feast M. W., 2015, MNRAS, 452, 910Miller A. A., Richards J. W., Bloom J. S., Cenko S. B., Silverman J. M.,Starr D. L., Stassun K. G., 2012, ApJ, 755, 98Orosz G., et al., 2017, AJ, 153, 119Pawlowski M. S., McGaugh S. S., 2014, MNRAS, 440, 908Pe˜na M., Stasi´nska G., Richer M. G., 2007, A&A, 476, 745P´erez-Mesa V., Zamora O., Garc´ıa-Hern´andez D. A., Plez B., ManchadoA., Karakas A. I., Lugaro M., 2017, A&A, 606, A20Pojmanski G., 1997, Acta Astron., 47, 467Riebel D., et al., 2015, ApJ, 807, 1Riechers D., Balega Y., Driebe T., Hofmann K. H., Men’shchikov A. B.,Shenavrin V. I., Weigelt G., 2005, A&A, 436, 925Samus’ N. N., Kazarovets E. V., Durlevich O. V., Kireeva N. N., PastukhovaE. N., 2017, Astronomy Reports, 61, 80Soszy´nski I., Gieren W., Pietrzy´nski G., Bresolin F., Kudritzki R.-P., StormJ., 2006, ApJ, 648, 375Soszy´nski I., et al., 2009, Acta Astron., 59, 239Soszy´nski I., et al., 2011, Acta Astron., 61, 217Suh K.-W., Kim H. Y., 2002, A&A, 391, 665Tramper F., Sana H., de Koter A., Kaper L., Ram´ırez-Agudelo O. H., 2014,A&A, 572, A36VanderPlas J. T., Ivezi´c ˇZ., 2015, ApJ, 812, 18Watson C. L., Henden A. A., Price A., 2006, Society for Astronomical Sci-ences Annual Symposium, 25, 47Weisz D. R., et al., 2011, ApJ, 739, 5Whitelock P. A., 2017, preprint, ( arXiv:1711.08932 )Whitelock P. A., Feast M. W., 2014, in EAS Publications Series. pp 263–269, doi:10.1051 / eas / arXiv:1807.03544 )Zuckerman B., Dyck H. M., 1986, ApJ, 304, 394 de Vaucouleurs G., de Vaucouleurs A., Corwin Jr. H. G., Buta R. J., PaturelG., Fouqu´e P., 1991, Third Reference Catalogue of Bright Galaxies.Volume I: Explanations and references. Volume II: Data for galaxiesbetween 0 h and 12 h . Volume III: Data for galaxies between 12 h and24 h .van Langevelde H. J., van der Heiden R., van Schooneveld C., 1990, A&A,239, 193van Loon J. T., Cohen M., Oliveira J. M., Matsuura M., McDonald I., SloanG. C., Wood P. R., Zijlstra A. A., 2008, A&A, 487, 1055 APPENDIX A:
GAIA
PHOTOMETRY AND CLASSES OFVARIABLES
Lebzelter et al. (2018) have proposed an interesting method forseparating AGB stars into di ff erent subgroups (e.g. C-rich, O-richAGB stars of low mass, etc) based on Gaia
DR2(Gaia Collabo-ration et al. 2016, 2018) BP and RP magnitudes combined with2MASS JK S magnitudes. They considered the LMC and showedhow stars grouped in the K S , W RP − W K plane, where W RP and W K are Wesenheit functions (see Lebzelter et al. 2018, for details).We constructed such a diagram for SMC AGB stars and founda very similar distribution to that for the LMC, in spite of the metal-licity di ff erence. Encouraged by this we computed the Wesenheitmagnitudes for the long-period variables in NGC 3109, IC 1613,NGC 6822, the three very long period SMC variables from Ta-ble 3 and the eight variables from Table A2 that have Gaia data.A matching radius of 1 arcsec was used in all cases.The data for the Miras are tabulated in Table A1 where BP and RP are Gaia
DR2 blue and red magnitudes, and W RP − W K are’colours’ as defined in Lebzelter et al. (2018). The supergiants andS stars are in Table A2. In Fig. A1 we plot M K S against W RP − W K .The solid lines separate the regions defined by Lebzelter et al.(2018); they are to some extent arbitrary and points lying close tothem cannot be confidently assigned to a particular region. We haveincluded known supergiants in the galaxies in the diagram. Abso-lute K S magnitude is plotted on the y axis in contract to apparentLMC K S magnitude in Lebzelter et al. (2018).Because the stars are relatively faint, Gaia photometry is notavailable for them all. In IC 1613 only the four brightest O-richMiras were found. These were considered by Menzies et al. (2015)to be massive stars undergoing HBB, and they appear in Fig. A1 inregion (b) as expected. All the unambiguous red supergiants clusterin region (d), while the Cepheid, v25, lies in region (c).For NGC 6822 there are data for seven of the 50 C-rich Miras(Whitelock et al. 2013) which all fall in region (b), and 10 of the 11O-rich Miras. The four HBB stars and the possible red supergiantappear in region (d), and two of the other O-rich variables are inregion (c). However, the remaining three are unexpectedly in region(b) amongst the C-rich stars. As discussed by Lebzelter et al. (2018)such stars might have dust shells which contribute extra reddeningnot accounted for in the Wesenheit formulae. The non-variable redsupergiants appear in region (d), while the S stars are to be foundin the lower C star region as Lebzelter et al. (2018) suggested theywould be.Six of the eight Miras in NGC 3109 reported in this paper havedata. Four of the suspected HBB stars fall in region (d) as would beexpected, while a fifth is in region (b), close to the boundary withregion (d). The very long period star (
MNRAS , 000–000 (0000) Menzies et al.
Table A1. JK S and Gaia ( BP , RP ) Photometry of LPV in LG GalaxiesGalaxy / Star * RA Dec K S J − K S RP BP − RP W RP − W K (2000.0)IC 16131017 16.27831 2.22648 15.14 1.11 18.26 2.53 0.603011 16.20428 2.09140 15.17 1.01 17.92 2.57 0.102035 16.11180 2.17299 15.20 1.13 17.99 2.15 0.771016 16.26576 2.19903 15.04 1.14 18.05 1.99 1.20NGC 682212557 296.18396 -14.78028 17.12 1.11 19.22 0.56 2.1420331 296.20413 -14.63472 15.46 1.12 17.97 2.26 0.3410184 296.22358 -14.77472 15.13 1.31 17.97 1.53 1.7530133 296.21812 -14.88028 15.09 1.21 18.36 1.88 1.6620134 296.21321 -14.68083 15.07 1.28 18.12 2.76 0.3540139 296.20425 -14.74278 13.92 1.33 17.29 2.71 0.7610198 296.25087 -14.76778 14.19 1.31 17.08 2.08 1.0930292 296.18800 -14.87222 14.68 1.20 17.79 2.68 0.4510091 296.26700 -14.76306 14.18 1.28 17.23 2.46 0.7320004 296.21292 -14.73222 12.57 1.15 15.92 2.91 0.3612790 296.20389 -14.75535 16.59 1.44 18.91 1.30 1.6210817 296.23639 -14.83054 16.04 1.66 18.38 0.79 2.4520540 296.18011 -14.71067 16.39 1.49 19.01 1.58 1.5940590 296.28567 -14.73952 16.39 1.62 19.16 1.50 1.9312751 296.21027 -14.75973 16.29 1.60 18.79 1.17 2.0820578 296.29065 -14.69641 16.14 1.66 18.98 1.09 2.5620542 296.17703 -14.71031 16.02 1.82 19.28 1.88 2.07NGC 31091067 150.73837 -26.14197 16.10 1.00 18.60 1.82 0.821112 150.80582 -26.15483 16.04 1.14 18.24 1.44 1.111153 150.75345 -26.17370 16.28 1.08 18.65 1.80 0.762075 150.91414 -26.17125 16.56 1.61 19.31 1.46 1.962081 150.86487 -26.15757 16.77 1.18 19.14 1.60 1.103064 150.65652 -26.14415 15.84 1.15 18.61 1.93 1.05SMCIRAS00483–7347 12.52963 -73.52377 8.71 2.83 16.19 3.67 4.65GM103 12.62764 -72.85830 8.91 1.70 13.34 4.94 -0.82HV11417 15.20071 -72.85056 8.86 1.44 14.34 4.81 0.21LMCIRAS 04407–7000 070.118667 -69.920417 8.69 1.96 15.45 5.84 0.51IRAS 04498–6842 072.422792 -68.630944 7.49 1.64 13.04 5.68 -0.69IRAS 04509–6922 072.668583 -69.292194 7.93 1.95 14.35 5.29 0.88IRAS 04516–6902 072.870792 -68.963889 7.91 2.01 15.70 4.14 3.79MSX LMC 1210 073.835417 -68.377583 9.72 2.94 16.44 3.25 4.51HV 888 076.058875 -67.270639 6.78 1.23 9.53 2.66 0.14MSX LMC 642 082.200708 -71.041361 9.88 3.12 14.74 5.18 0.27IRAS 05558–7000 088.837625 -70.000833 9.13 2.86 17.60 4.41 4.69 * Distance moduli used in Fig. A1: IC 1613 (24.4), NGC 6822 (23.5), NGC 3109 (25.6), SMC (18.92), LMC(18.50). pected. In the last case, there must be a very thick dust shell pro-viding extra reddening.For the LMC stars, we only have single-epoch 2MASS mea-surements. These stars are likely to have relatively large amplitudesso the positions in Fig. A1 in the vertical direction are somewhatmore uncertain than for the stars in the other galaxies. Nevertheless,three of the LMC stars lie in a similar position to the single ex-tremely red SMC Mira and, like it, probably have thick dust shells.None of the WLM variables have
Gaia
DR2 data so could notbe included.Thus, most of this small sample of Miras in Local Groupgalaxies appear in the Lebzelter diagram where predicted on thebasis of the LMC data. However, this diagram can only be used assupporting evidence for the chemical characteristics of a specificvariable; spectra are necessary for a definite attribution to be made.
APPENDIX B: VERY LONG PERIOD VARIABLES IN THESMC
The measurements reported here were obtained with the 1.9-m tele-scope at SAAO and are on the SAAO photometric system as de-fined by Carter (1990), i.e. they are di ff erent from the IRSF pho-tometry reported in the body of the paper (Carpenter (2001) pro-vides transformation equations between the SAAO and 2MASSsystems). The errors are less than 0.03 mag at JHK and less than0.05 mag at L except where marked with a colon where they are lessthan 0.1 mag. Fourier mean values for the photometry are given inTable 3. MNRAS000
The measurements reported here were obtained with the 1.9-m tele-scope at SAAO and are on the SAAO photometric system as de-fined by Carter (1990), i.e. they are di ff erent from the IRSF pho-tometry reported in the body of the paper (Carpenter (2001) pro-vides transformation equations between the SAAO and 2MASSsystems). The errors are less than 0.03 mag at JHK and less than0.05 mag at L except where marked with a colon where they are lessthan 0.1 mag. Fourier mean values for the photometry are given inTable 3. MNRAS000 , 000–000 (0000) uminous AGB variables in NGC 3109 Table A2. JK S and Gaia ( BP , RP ) Photometry of supergiants and S stars in LG GalaxiesGalaxy / Star * RA Dec K S J − K S RP BP − RP W RP − W K (2000.0)IC 1613 supergiants1027 16.25295 2.18007 15.52 0.51 16.85 1.18 0.154013 16.22231 2.08931 14.60 0.87 16.69 1.81 0.341004 16.24328 2.15233 13.19 0.91 15.80 2.40 0.111003 16.25696 2.14422 13.01 0.92 15.29 1.90 0.441009 16.27144 2.19849 14.31 0.82 16.29 1.68 0.361010 16.29004 2.20715 13.98 0.92 16.16 1.83 0.421008 16.31963 2.18765 13.92 0.91 16.17 1.95 0.343003 16.16047 2.01605 13.34 0.82 15.44 2.05 -0.01NGC 6822 supergiants30016 296.19067 -14.87276 12.78 1.13 15.56 2.25 0.6340115 296.19919 -14.84817 13.26 1.15 16.08 2.37 0.5240177 296.21021 -14.73628 14.06 1.08 16.63 2.01 0.7010089 296.22278 -14.76476 14.02 1.08 16.50 1.88 0.7810032 296.22696 -14.80191 13.34 1.14 16.07 2.25 0.5940278 296.22726 -14.85778 12.33 1.10 15.08 2.27 0.5540315 296.23212 -14.86564 12.33 1.06 15.11 2.38 0.4110011 296.23883 -14.82247 12.46 1.14 15.29 2.26 0.6710015 296.24945 -14.75443 12.70 0.94 14.99 1.96 0.3920101 296.26492 -14.72723 14.81 1.15 17.38 1.88 0.91NGC 6822 S stars10870 296.17892 -14.82286 16.19 1.34 18.98 1.44 1.8410784 296.21545 -14.83469 16.20 1.26 18.53 1.24 1.5811004 296.27341 -14.80861 16.27 1.33 18.77 1.12 1.9711029 296.28308 -14.80497 16.22 1.24 18.66 1.00 1.9930528 296.19156 -14.89296 16.51 1.26 19.27 1.60 1.5510326 296.25522 -14.82579 15.52 1.30 18.21 1.57 1.54 * Distance moduli used in Fig. A1: IC 1613 (24.4), NGC 6822 (23.5).
Table B1.
HV11417JD
J H K L − APPENDIX C: PHOTOMETRIC DATA FOR NGC 3109
The time-series photometry for the variables reported on in thispaper is presented in the following nine tables, Tables C1 to C9.Column contents are indicated by the headings and are obvious ex-cept that (cid:15) J , etc are photometric errors. A sample of the photometricdata catalogue is shown in the accompanying Table C10. The fullphotometric catalogue and the variable star files can be obtainedonline. Table B2.
IRAS00483–7347JD
J H K L − , 000–000 (0000) Menzies et al.
Figure A1.
Modified Lebzelter et al. (2018) diagram for long-period variables in NGC 3109 ( ∗ ), IC 1613 ( + ) including supergiants ( (cid:52) ), NGC 6822 (x)including supergiants ( ◦ ) and S stars ( (cid:53) ), three very long period variables in the SMC ( (cid:5) ) and eight very long period O-rich variables in the LMC (thin (cid:5) ). Alarge circle is placed around the symbols for the known or suspected C stars for emphasis. The annotated regions indicate where stars of a particular type arefound, viz., (a) low mass O-rich AGB stars, (b) C-rich stars, (c) O-rich AGB stars of intermediate mass, (d) red supergiants and O-rich massive AGB stars, and(e) RGB and faint AGB stars. Absolute K S magnitude is plotted on the y axis using ( m − M ) = . Table B3.
GM103JD
J H K L − Table C1.
Photometric data for variable J (cid:15) J H (cid:15) H K S (cid:15) K000
Photometric data for variable J (cid:15) J H (cid:15) H K S (cid:15) K000 , 000–000 (0000) uminous AGB variables in NGC 3109 Table C2.
Photometric data for variable J (cid:15) J H (cid:15) H K S (cid:15) K Table C3.
Photometric data for variable J (cid:15) J H (cid:15) H K S (cid:15) K Table C4.
Photometric data for variable J (cid:15) J H (cid:15) H K S (cid:15) K Table C5.
Photometric data for variable J (cid:15) J H (cid:15) H K S (cid:15) K Table C6.
Photometric data for variable J (cid:15) J H (cid:15) H K S (cid:15) K Table C7.
Photometric data for variable J (cid:15) J H (cid:15) H K S (cid:15) K , 000–000 (0000) Menzies et al.
Table C8.
Photometric data for variable J (cid:15) J H (cid:15) H K S (cid:15) K Table C9.
Photometric data for variable J (cid:15) J H (cid:15) H K S (cid:15) K Table C10.
Photometric Data for NGC 3109RA Dec N a J σ J b H σ H b K S σ K b J − H H − K S J − K S Note c (J2000.0)150.75345 -26.17370 1153 17.367 0.379 16.682 0.405 16.284 0.371 0.685 0.398 1.083 V150.80502 -26.16340 1171 18.258 0.062 17.503 0.104 17.463 0.284 0.755 0.040 0.795150.81224 -26.16286 1173 18.018 0.110 17.109 0.079 16.704 0.125 0.909 0.405 1.314 C150.79617 -26.16277 1174 17.823 0.061 17.080 0.046 16.973 0.116 0.743 0.107 0.850150.80377 -26.16170 1177 18.196 0.145 17.310 0.123 16.752 0.110 0.886 0.558 1.444 G a catalogue star number b standard deviation of tabulated magnitude c V indicates variable ; C indicates
C star ; G indicates probable background galaxy
MNRAS000