The 6-GHz multibeam maser survey III: comparison between the MMB and HOPS
S. L. Breen, Y. Contreras, S. P. Ellingsen, J. A. Green, A. J. Walsh, A. Avison, S. N. Longmore, G. A. Fuller, M. A. Voronkov, J. Horton, A. Kroon
MMon. Not. R. Astron. Soc. , 000–000 (0000) Printed 21 July 2018 (MN L A TEX style file v2.2)
The 6-GHz multibeam maser survey III: comparisonbetween the MMB and HOPS
S. L. Breen, (cid:63) , Y. Contreras, S. P. Ellingsen, J. A. Green, A. J. Walsh, A. Avison, , S. N. Longmore, G. A. Fuller, , M. A. Voronkov, J. Horton, A. Kroon Sydney Institute for Astronomy (SIfA), School of Physics, University of Sydney, NSW 2006, Australia; Leiden Observatory, Leiden University, P.O. Box 9513, NL-2300 RA Leiden, The Netherlands; School of Mathematics and Physics, University of Tasmania, Private Bag 37, Hobart, Tasmania 7001, Australia; CSIRO Astronomy and Space Science, Australia Telescope National Facility, PO Box 76, Epping, NSW 1710, Australia; International Centre for Radio Astronomy Research, Curtin University, GPO Box U1987, Perth, WA 6845, Australia; Jodrell Bank Centre for Astrophysics, Alan Turing Building, School of Physics and Astronomy, The University of Manchester,Manchester M13 9PL, UK; UK ALMA Regional Centre Node, Manchester, M13 9PL, UK; Astrophysics Research Institute, Liverpool John Moores University, 146 Brownlow Hill, Liverpool L3 5RF, UK
21 July 2018
ABSTRACT
We have compared the occurrence of 6.7-GHz and 12.2-GHz methanol masers with22-GHz water masers and 6035-MHz excited-state OH masers in the 100 square degreeregion of the southern Galactic plane common to the Methanol Multibeam (MMB)and H O southern Galactic Plane surveys (HOPS). We find the most populous starformation species to be 6.7-GHz methanol, followed by water, then 12.2-GHz and,finally, excited-state OH masers. We present association statistics, flux density (andluminosity where appropriate) and velocity range distributions across the largest, fullysurveyed portion of the Galactic plane for four of the most common types of masersfound in the vicinity of star formation regions.Comparison of the occurrence of the four maser types with far-infrared dust tem-peratures shows that sources exhibiting excited-state OH maser emission are warmerthan sources showing any of the other three maser types. We further find that sourcesexhibiting both 6.7-GHz and 12.2-GHz methanol masers are warmer than sources ex-hibiting just 6.7-GHz methanol maser emission. These findings are consistent withpreviously made suggestions that both OH and 12.2-GHz methanol masers generallytrace a later stage of star formation compared to other common maser types.
Key words: masers – stars: formation – ISM: molecules – radio lines: ISM
Masers are important probes of a number of types of as-tronomical objects, and are particularly prevalent towardsregions of high-mass star formation. Two of the most com-monly detected masers in our Galaxy arise from the 6.7-GHzmethanol and 22-GHz water transitions (e.g. Breen et al.2015; Walsh et al. 2014), followed by a host of other transi-tions that include the 12.2-GHz methanol maser (e.g. Breenet al. 2016) and 6035-MHz excited OH masers (e.g. Avi-son et al. 2016). Other, rarer, maser transitions also havea special role to play - signposting physical conditions that (cid:63)
Email: [email protected] are less commonly found, or, perhaps more likely, associatedwith short-lived evolutionary phases in the star formationprocess (e.g. Ellingsen et al. 2011, 2013).The exclusive association between 6.7-GHz methanolmasers and young high-mass stars (e.g. Minier et al. 2003;Xu et al. 2008; Breen et al. 2013) make them particularlyuseful for pinpointing and studying high-mass star forma-tion. Since they also have a tendency to trace systemic ve-locities (e.g. Szymczak et al. 2007; Caswell 2009; Pandianet al. 2009; Green & McClure-Griffiths 2011) they have beensuccessfully used to trace Galactic structure (Green et al.2011). They are relatively common and strong, with morethan 1000 sources now known across the Galactic plane (e.g.Pandian et al. 2007; Caswell et al. 2010; Green et al. 2010; c (cid:13) a r X i v : . [ a s t r o - ph . GA ] N ov S. L. Breen et al.
Caswell et al. 2011; Green et al. 2012; Breen et al. 2015).The next strongest and common class II methanol maserline is at 12.2-GHz (e.g. Caswell et al. 1995; B(cid:32)laszkiewicz &Kus 2004; Gaylard, MacLeod & van der Walt 1994; Breenet al. 2012a,b, 2014, 2016).Methanol masers at 12.2-GHz have always been foundto have 6.7-GHz counterparts, and although a large, sensi-tive, unbiased survey for 12.2-GHz methanol maser emissionhas never been conducted, we can be relatively certain thata near-complete sample of these masers can be gained bytargeting a complete sample of 6.7-GHz masers. While inrecent years most 12.2-GHz methanol maser searches havebeen targeted towards 6.7-GHz methanol masers, some ofthe earlier searches targeted known sites of OH maser emis-sion (e.g. Kemball, Gaylard & Nicolson 1988; Caswell et al.1993), water maser emission (e.g. Koo et al. 1988) and othersources indicative of star formation regions (e.g. Norris et al.1987). All of these early 12.2-GHz detections were later re-ported as sources of 6.7-GHz methanol maser emission, andthis, combined with the fact that there has never been aserendipitous detection of 12.2-GHz methanol maser emis-sion without a 6.7-GHz counterpart, and the fact that 12.2-GHz methanol maser emission only rarely shows peak fluxdensities that surpass that of their 6.7-GHz counterpart (e.g.Breen et al. 2011) clearly indicates that there is not a sig-nificant population of 12.2-GHz methanol masers devoid of6.7-GHz counterparts.The two transitions are typically co-spatial to within afew milliarcseconds (e.g. Norris et al. 1993; Moscadelli et al.2002) and therefore can be used in combination to reveal thephysical conditions in the star formation regions they aredetected towards, as the conditions required to produce thetwo transitions are similar, but not identical (Cragg, Sobolev& Godfrey 2005). This means that the presence or absenceof accompanying 12.2-GHz emission may be determined byonly a small change in physical conditions, which has beenproposed to be a reflection of the evolutionary stage of theassociated high-mass star formation region (e.g. Ellingsenet al. 2007; Breen et al. 2010a).Galactic water masers have been detected towards re-gions of star formation with masses extending to low-massobjects (e.g. Claussen et al. 1996; Furuya et al. 2001) as wellas evolved stars (e.g. Deacon et al. 2007). Water masers canexhibit extreme levels of temporal variability (e.g. Brandet al. 2003; Felli et al. 2007), meaning that single-epochsearches can misrepresent true detection statistics. Com-bined, the tendency for water masers to be associated with arange of exciting objects and their temporal variability havehampered efforts to study a complete population of watermasers associated with high-mass stars.The relative occurrence of 6.7-GHz methanol masersand water masers has been previously investigated throughtargeted observations (e.g. Beuther et al. 2002; Szymczaket al. 2005; Xu et al. 2008; Breen et al. 2010b; Titmarshet al. 2014, 2016). Three of these significant investigations(Szymczak et al. 2005; Titmarsh et al. 2014, 2016) targetedtheir water maser observations towards statistically com-plete samples of 6.7-GHz methanol masers, finding watermaser detection rates of ∼
50 per cent. Xu et al. (2008) made6.7-GHz methanol maser observations towards known 22-GHz water masers, yielding a detection rate of ∼
35 per cent(once low-mass objects were excluded). Beuther et al. (2002) targeted their 6.7-GHz methanol and water maser observa-tions towards 29 high-mass star forming regions, finding that ∼
38 per cent of methanol masers show no water maser emis-sion and ∼
35 per cent of water masers show no methanolmaser emission. Breen et al. (2010b) conducted a sensitiveAustralia Telescope Compact Array (ATCA) search of 2706.7-GHz methanol masers, finding a water maser detectionrate of 73 per cent. In these previous studies comparing6.7-GHz and water maser occurrence, biases in the way thetargets were selected, and indeed the fact that the observa-tions were targeted at all, have heavily affected the detectionstatistics, resulting in association rates between 35 and 73per cent.Likewise, until the MMB survey the detection rate of12.2-GHz methanol maser emission towards 6.7-GHz sourcesvaried from 19 to 60 percent (Gaylard, MacLeod & van derWalt 1994; Caswell et al. 1995; B(cid:32)laszkiewicz & Kus 2004;Breen et al. 2010b) depending on the sample selection andthe sensitivity of the observations. Since the 12.2-GHz searchthat targeted the MMB 6.7-GHz maser sample, we now un-derstand that the true detection rate is ∼
45 per cent (Breenet al. 2016). Studies of large, complete samples of excited-state OH masers are much rarer, perhaps limited to theMMB itself (Avison et al. 2016).Following the completion of both the Methanol Multi-beam (MMB) Survey (Green et al. 2009) and the H OGalactic Plane Survey (HOPS; Walsh et al. 2011) we are nowin a position to conduct a definitive comparison between sta-tistically complete sample of the most common masers foundtowards high-mass star forming regions. Here we present acomparison of the occurrence of 6.7-GHz methanol masers,12.2-GHz methanol masers, water masers and 6035-MHzexcited-OH masers located within a 100 square degree por-tion of the southern Galactic plane made possible by theMMB and HOPS surveys described in Sections 1.1 and 1.2.
The MMB survey searched the portion of the SouthernGalactic plane visible to the Parkes 64-m radio telescope.While the primary target of the survey was the 6.7-GHzmethanol maser line, a concurrent survey for 6035-MHz OHmasers was also preformed (see Green et al. 2009, for a de-scription of the survey or Table 1 for a summary of thesurvey parameters). In total, 972 methanol masers and 127excited-state OH masers were detected and are presented ina series of catalogue papers (Caswell et al. 2010; Green et al.2010; Caswell et al. 2011; Green et al. 2012; Breen et al. 2015;Avison et al. 2016). Each of the newly discovered 6.7-GHzmethanol and excited-state OH masers, together with anyknown sources that had not been previously observed withan interferometer where subsequently observed with eitherthe ATCA or MERLIN between 2006 and 2014, achievingpositional accuracies better than 0.4 arcsec. Each of the de-tected masers was re-observed with Parkes in a sensitive‘MX’ observation (where the on-source position is cycledthrough each of the seven beams of the receiver) that hadan rms noise of ∼ c (cid:13) , 000–000 MB versus HOPS O southern Galactic Plane Survey(HOPS)
HOPS observed 100 square degrees of the Southern Galacticplane for a number of spectral lines between 19.5 and 27.5GHz, including water masers (Walsh et al. 2011). Observa-tions were made outside the traditional mm season whichresulted in a slightly variable detection limit across the sur-vey region. All water maser detections were followed up withmore sensitive ATCA observations to derive precise posi-tions (Walsh et al. 2014). See Table 1 for a summary of thesurvey parameters.A total of 2790 water maser spots were detected, dis-tributed amongst 631 maser sites (defined by Walsh et al.(2014) to be any spots enclosed within a 4 arcsec radius).A total of 31 masers detected in the initial survey were notdetected during the ATCA follow up observations (consis-tent with expectations of water maser variability), meaningthat 122 additional sites were detected in the ATCA obser-vations. These additional sources can easily be accountedfor by the vast improvements in both spatial resolution andsensitivity.Of the 631 water masers detected by Walsh et al. (2014),121 (19 per cent) are associated with evolved stars and aretherefore excluded from our sample. Walsh et al. (2014) alsodesignated 77 (12 per cent) of their water maser sources asunknown but we found that two of these are associated withMMB sources and have therefore changed their designationto reflect that they are associated with star formation re-gions. In the following analysis we consider only the 435water masers found to be associated with star formation re-gions. The Walsh et al. (2014) star formation designationsdo not discriminate on the basis of mass range but the vastmajority of the water masers designated as being associatedwith star formation here will be associated with high-massstar formation. If HOPS has detected water masers towardslow mass stars, they would fall into the unknown categorysince the star formation designations rely on the associationwith 6.7-GHz methanol masers or infrared characteristicsindicative of high-mass star formation.
Determining associations between water and other maserscan be difficult, due to the intrinsic difference in spot dis-tributions, site definitions and positional uncertainties. Thespots associated with a methanol maser site tend to be lo-cated within a fairly compact region, typically containedwithin less than 2 arcsec (Caswell 2009; Caswell et al. 2010) while water maser emission can be spread over 4 arcsec (cor-responding to corresponding to 0.1 pc at a distance of 5 kpc)or more (e.g. Reid et al. 1988). The tendency for water maseremission to be more scattered is due to their collisional ex-citation, allowing spots to be associated with outflows, forexample, and causing them to be spread over greater ex-tents. This presents a challenge when assigning individualspots to a water maser site, especially given the large popu-lations of water masers and crowded star formation regions.To combat this, HOPS reported individual maser spots aswell as categorising spots into sites, which were defined byan upper limit on the spot distribution of 4 arcsec. The posi-tional uncertainties are generally much less of a concern: theuncertainty of the MMB survey is 0.4 arcsec (e.g. Caswellet al. 2010) and the water maser positions from HOPS areexpected to be accurate to around one arcsec.To account for the differences in spot distributionsand site definitions, we have compared the MMB methanolmaser site positions with the HOPS spot positions in Fig. 1.This comparison should allow us to capture more true asso-ciations as the separations will not be dominated by largewater maser spreads since we are considering every spot in-dividually. There are 218 MMB sources that have a HOPSwater maser spot within 20 arcsec. Fig. 1 shows that themajority of these (183) are separated by ≤ Within the HOPS survey range we can compare the detec-tion statistics of complete surveys for 6.7-GHz and 12.2-GHzmethanol masers, 6035-MHz excited-state OH masers andwater masers. Table 2 shows that within the 100 square de-grees, there are 634 6.7-GHz methanol masers, 435 22-GHzwater masers associated with star formation, 295 12.2-GHzmethanol masers and 80 excited-state OH masers. As dis-cussed in Section 3.1, the lower sensitivity of the HOPS sur-vey results in an underestimation of the number of watermasers, and if that survey had a detection limit comparableto the other searches, it is likely that it would have foundhundreds more sources in the survey range (perhaps exceed-ing the number of 6.7-GHz sources).Table 2 gives the number of masers of each transition c (cid:13) , 000–000 S. L. Breen et al.
Table 1.
Survey parameters of the surveys described in Sections 1.1 and 1.2. Column 1 indicates the survey name, column 2 gives thetelescopes used in those surveys (here P = Parkes, A = ATCA, M = MERLIN, Mop = Mopra). Column 3 gives the maser targeted in thesurvey, column 4 gives the Galactic coverage of the survey, column 5 gives the year of the initial survey observations in the 2000s, column6 gives the typical 5- σ values of each of the surveys, column 7 gives the values at which the MMB and HOPS surveys are estimated to becomplete at, column 8 gives the velocity resolution of the survey data, and column 9 gives the years that the data used in the analysiswas chiefly collected in (since for all but the 12.2-GHz survey, it is subsequent, more sensitive follow-up data used in this analysis). Survey Telescope Maser Coverage Initial 5- σ Complete Vel res Epochsurvey (Jy) (Jy) (km s − ) (yr)(yr) MMB P, A, M 6.7-GHz meth 186 ◦ 06 - 09 0.85 1 0.11 06 - 09P, A 6035-MHz OH 186 ◦ 06 - 09 0.85 0.12 06 - 0912.2-GHz P 12.2-GHz meth MMB targeted 08, 10, 15 0.85 0.08 08, 10, 15HOPS Mop, A 22-GHz water 290 ◦ 08 - 10 2 10 0.42 11, 12 Separation (arcsec) N u m be r o f s ou r c e s Figure 1. Separation between MMB source and nearest watermaser spot reported in HOPS within a 20 arcsec threshold. detected in the 100 square degree region and Fig. 2 showsthe associations between the species. We find that 259 of thestar formation water masers are devoid of methanol masercounterparts, meaning that 40 per cent of star formation wa-ter masers have a methanol maser counterpart. The comple-mentary statistic is that 28 per cent of the 6.7-GHz methanolmasers have associated water masers detected in HOPS.The percentage of 6.7-GHz methanol masers that haveassociated water maser emission is approximately the samewhether or not the 6.7-GHz emission is accompanied by12.2-GHz emission (31 per cent) or not (27 per cent). Breenet al. (2014) compared the occurrence of water masersand 12.2-GHz methanol masers in the 6 ◦ to 20 ◦ longituderange, finding that water masers were more often found to-wards 6.7-GHz methanol masers with accompanying 12.2-GHz methanol maser emission (detection rate of 55.7 per-cent compared to 36.2 percent for 6.7-GHz methanol masersdevoid of 12.2-GHz emission). The main difference betweenthe samples is a vastly different water maser detection limit,suggesting that the less sensitive HOPS sample is missing a Table 2. Detection statistics of the different maser species withinthe HOPS survey range.Maser type higher relative number of water masers that are associatedwith both 6.7-GHz and 12.2-GHz sources. In their obser-vations of water masers towards 6.7-GHz methanol masers,Titmarsh et al. (2016) detected a smaller proportion of wa-ter masers with flux densities less than 2 Jy towards 6.7-GHzmethanol masers devoid of accompanying 12.2-GHz maseremission (34 per cent) compared to those sources with both6.7- and 12.2-GHz emission (47 per cent). Although the sam-ple size is inadequate to draw firm conclusions (34 watermasers accompanying both 6.7- and 12.2-GHz emission and67 associated with just 6.7-GHz methanol maser emission)this is consistent with relatively more water masers associ-ated with both 6.7- and 12.2-GHz falling below the detectionlimit of the HOPS observations.Just over half (54 per cent) of the 80 excited-stateOH masers are associated with 6.7-GHz methanol masers,and are more commonly associated with 6.7-GHz methanolmasers that also have 12.2-GHz emission (27 compared to16) or sources that have any combination of 2 or more othermasers (36 compared to 14). This is consistent with thenotion that OH masers are associated with a slightly laterphase of the star formation process (e.g. Caswell 1997). While the maser surveys we are comparing are all completeover the same range of Galactic longitude, the HOPS sur-vey has a much higher detection limit than the others whichmeans that the number of water masers (and therefore wa-ter maser associations) should be considered a lower limit.Even with the higher detection limit, HOPS is still a valu-able resource, being the largest systematic search for watermaser emission ever conducted. It is also worth noting thateven though the searches for excited-state OH, 6.7-GHz and c (cid:13) , 000–000 MB versus HOPS Figure 2. Venn diagram showing the associations between thefour maser species: 6.7-GHz and 12.2-GHz methanol, excited-state OH and water. Note that five water masers have spots areassociated with two 6.7-GHz methanol masers, and one watermaser has spots associated with three 6.7-GHz methanol masers(and some of these are also associated with 12.2-GHz methanolor excited-state OH emission) and therefore the number of watermasers shown in this diagram is 442 (i.e. seven more than thetotal number of water masers). ◦ (through the GC)to 20 ◦ . The detection limit of this survey was ∼ 40 timeslower than the completeness level of the HOPS survey andhas therefore detected a higher fraction of the water maserpopulation towards 6.7-GHz methanol masers. Our detec-tion statistics show that 28 per cent of 6.7-GHz methanolmasers have an associated HOPS water maser. This impliesthat HOPS was able to recover about 58 per cent of thewater masers towards a sample of 6.7-GHz methanol masersthat a more sensitive survey would. It is difficult to estimatehow many solitary water masers were missed by HOPS sincethere is no large scale, sensitive survey available for compari-son. If we assume that the luminosity distribution of solitarywater masers is similar to the distribution of water masersassociated with 6.7-GHz methanol masers then we wouldalso expect that ∼ 42 per cent of sources are missing. How-ever, previous studies have shown that solitary water maserstend to have lower peak flux densities than those associatedwith 6.7-GHz methanol masers (Breen et al. 2010b; Breen &Ellingsen 2011), so the sensitivity limitations of HOPS will have prevented an even higher percentage of these solitarywater masers from being detected.A further difficulty presented by water maser detectionstatistics is that a single epoch search will not recover the en-tire population of sources due to their high level of temporalvariability. HOPS itself is a good example of this; the initialsurvey with the Mopra radio telescope detected 540 watermaser sites (Walsh et al. 2011) and 31 of those sources werenot detected in the much more sensitive ATCA follow-up ob-servations. This indicates that ∼ ∼ Fig. 3 shows box plots of the velocity ranges of each of the6.7-GHz methanol, 12.2-GHz methanol, excited-state OHand water masers. The excited-state OH maser data pre-sented in Avison et al. (2016) gives values for LHCP (lefthand circularly polarised) and RHCP (right hand circu-larly polarised) emission separately and we have used the c (cid:13) , 000–000 S. L. Breen et al. Velocity range (km/s) w a t e r e x O H . Figure 3. Velocity range box plots of each of the four masertypes; 6.7-GHz methanol (magenta), 12.2-GHz methanol (pink),excited-state OH (orange) and water (blue). Water masers withvelocity ranges in excess of 105 km s − (26; with ranges up to299.6 km s − ) have been excluded from the plot in order to see theother, smaller ranges more clearly. In each box plot, the solid ver-tical black line represents the median of the data, the coloured boxrepresents the interquartile range (25th to the 75th percentile),and the dashed horizontal lines (the ‘whiskers’) show the rangefrom the 25th percentile to the minimum value and the 75th per-centile to the maximum value, respectively. Values that fall morethe 1.5 times the interquartile range from either the 25th or 75thpercentile are considered to be outliers are represented by opencircles. larger of the LHCP or RHCP velocity range in this anal-ysis. The median velocity range is highest for the watermasers at 9.8 km s − , followed by the 6.7-GHz methanol(6.1 km s − ), excited-state OH (4.5 km s − ), and 12.2-GHzmethanol masers (1.9 km s − ). Across the full MMB range,the median velocity of 6.7-GHz methanol masers is 6 km s − (Green et al. 2017) and the median of 12.2-GHz sources is1.7 km s − (Breen et al. 2016) and therefore are consistentwith the sub-sample we consider here. The fact that thewater masers show the largest velocity range is expected,particularly because of their tendency to trace high-velocityoutflow, but the median velocity of the HOPS sources issignificantly lower than either of the Breen et al. (2010b)or Titmarsh et al. (2014, 2016) targeted water maser obser-vations which have medians of 15 and 17 km s − , respec-tively. A part of this difference can be accounted for by thefact that the Walsh et al. (2014) quoted peak velocities ofspots and therefore results in an underestimation of the ve-locity ranges of sites, but some of the difference is due tothe unbiased nature of HOPS. The fact that the velocityrange of the excited-state OH masers exceed the 12.2-GHzmethanol masers is perhaps surprising given that 12.2-GHzmethanol masers generally have much higher peak flux den-sities, and suggests that the excited-OH emission is arisingfrom a larger volume of gas than the 12.2-GHz maser emis-sion.Breen et al. (2016) found that there was a preponder-ance of 12.2-GHz methanol masers with high velocity rangeswithin the Galactic longitude ranges of G 330 - G 340 and toa slightly lesser extent, G10 - G30. These Galactic longitude ranges correspond to significant structures in the Galaxy andit was suggested that the higher velocity ranges reflected theenhanced star formation resulting in larger relative motionsthat were then reflected in the individual star forming re-gions. We have repeated this analysis using the HOPS watermasers, and found a similar peak in the G330 - G340 longi-tude range, but in the case of the water masers, this peakis entirely consistent with the higher number of sources inthis longitude range (i.e. ∼ 17 per cent of the water maserpopulation is located in this longitude range, and ∼ 19 percent of the population of water masers with velocity rangesin excess of the median water velocity range lie in this lon-gitude range). The fact that the velocity range of the watermasers do not show a significant overabundance of high ve-locity sources (as is the case for 12.2-GHz methanol masers)is likely due to the fact that water maser velocities are notas closely tied to the star formation region, instead oftendepending on the interaction of the young star formationregion with the surrounding environment. Future compar-isons with dense gas line-widths of star formation regions asa function of Galactic longitude will be able to shed light onthis peculiar result.Fig. 4 shows the velocity ranges of each of the four typesof masers split into their association categories of solitary,associated (with any combination of the other three masertypes), and associated with each of 6.7-GHz methanol, 12.2-GHz methanol, excited-state OH and water maser emission.In the case of 12.2-GHz methanol masers there are no ‘soli-tary’ sources since all have 6.7-GHz counterparts so for thesesources ‘associated’ means that they have accompanyingemission from either or both of excited-state OH and wa-ter maser emission. Immediately evident in Fig. 4 is the factthat ‘solitary’ masers have velocity range distributions thatare skewed towards smaller values. This has been noted pre-viously in the case of water masers (e.g. Breen & Ellingsen2011) and for 6.7-GHz methanol masers (e.g. Breen et al.2011).Since Walsh et al. (2014) published the peak velocityof each detected maser feature, masers consisting of a singlespectral feature have a calculated velocity range of zero. Itcan be seen in Fig. 4 that sources with velocity ranges of zerokm s − account for a large portion of the ‘solitary’ watermasers - in fact, there are 89 (equating to more than a thirdof sources in that category), compared to 26 ( ∼ 14 per cent)in the associated category. Breen et al. (2010b) found thatwater masers with single features, and those that have no6.7-GHz methanol or OH maser counterpart, are more likelyto fall below the detection limit of a sensitive, single-epochsearch and so it is likely that this population is particularlyunderrepresented in the HOPS sample.Walsh et al. (2014) presented their water masers as a se-ries of water maser spots. One might expect that the numberof water maser spots is closely linked to the velocity rangeof the maser emission and a comparison of Fig. 5, whichshows the number of water maser spots as a function of as-sociations, with the water maser velocity plot shown in thebottom right of Fig. 4, reveals that our data are broadlyconsistent with that assertion. In particular, it can be seenthat the water masers that are associated with excited-stateOH masers have the largest velocity ranges and also tend tohave higher numbers of spots, while solitary water masers c (cid:13) , 000–000 MB versus HOPS Velocity range (km/s) w a t e r e x O H ss o cs o l Velocity range (km/s) w a t e r e x O H a ss o c . Velocity range (km/s) w a t e r . ss o cs o l Excited−state OH Velocity range (km/s) e x O H . ss o cs o l water Figure 4. Velocity range box plots of 6.7-GHz methanol masers (top left), 12.2GHz methanol masers (top right), excited-state OHmasers (bottom left) and water masers (bottom right) in the categories of solitary (‘sol’; grey), associated with one or more of the othertransitions (‘assoc’; cyan), associated with 6.7-GHz methanol (‘6.7’; magenta), 12.2-GHz (‘12’; pink), excited-state OH (‘exOH’; orange)and water masers (‘water’; blue). Note that in the case of 12.2-GHz methanol masers ‘assoc’ includes those 12.2-GHz sources that areassociated with either or both of excited-state OH or water maser emission since all are associated with 6.7-GHz methanol maser sources.The water maser plot has been truncated in order to see detail in the other distributions and the full range of water maser outliersextends to 299.6 km s − (i.e. 11 water maser sources have been excluded from the plot). See Fig. 3 caption for a general explanation ofbox plots. have the lowest velocity ranges and the smallest number ofmaser spots. Methanol masers at 6.7-GHz are considered good probesof systemic velocities, with their central velocities most of-ten falling within ± − of the systemic velocities ofthe regions they are associated with (e.g. Szymczak et al.2007; Caswell 2009; Pandian et al. 2009; Green & McClure-Griffiths 2011). Water masers, on the other hand, are gen-erally considered to be poor tracers of systemic velocitiesdue to their association with outflows. We have comparedthe velocity of the peak water maser emission to the cen-tral 6.7-GHz methanol maser emission and find that ∼ ± 10 km s − of the central 6.7-GHz methanol maser veloc-ity. Of the 11 per cent of sources with velocities at largerseparations, we find that 15 sources are more blue-shiftedand 6 sources are more red-shifted. Similarly, in the watermaser observations targeting MMB sources, Titmarsh et al. (2016) found that 88 per cent of water masers showed theirpeak emission within ± 10 km s − of the methanol maserpeak, while in observations targeting large, but incompletesamples of 6.7-GHz methanol and OH masers, Breen et al.(2010b) found that 78 per cent of their sources shared peakvelocities within ± 10 km s − of each other. The consistencywith the Titmarsh et al. (2016) sample is probably reflec-tive of the fact that they also use a portion of the completesample of MMB masers, but it is interesting that the per-centage of sources remains constant with the much moresensitive Titmarsh et al. (2016) sample, suggesting that thedistribution holds to lower water maser peak flux densities.Considering that the central velocity (the middle ofthe maximum and minimum velocity) of 6.7-GHz methanolmasers is a more accurate tracer of systemic velocities thanthe velocity of the peak emission (e.g. Green & McClure-Griffiths 2011), we repeated the comparison using watermaser central velocity. In this case ∼ 74 per cent of the wa-ter maser central velocities fell within ± 10 km s − of thecentral 6.7-GHz velocity and of the 47 sources with greaterseparations, 29 were more blue-shifted and 18 were morered-shifted. So we conclude that the water maser central ve- c (cid:13) , 000–000 S. L. Breen et al. Number of water maser spots e x O H . ss o cs o l Figure 5. Box plots of the number of water maser spots as-sociated with water maser sites in the categories of solitary(‘sol’; grey), associated (‘assoc’; blue), associated with: 6.7-GHzmethanol masers (6.7; magenta), 12.2-GHz methanol masers (12;pink) and excited-state OH (exOH; orange). Here ‘solitary’ meansthose water masers that are not associated with any of the otherthree maser types and ‘associated’ means those water masersthat are associated with at least one of the other three masertypes. The plot range excludes one source that is associated witha methanol maser (G000.677-0.028, with 61 spots). See Fig. 3caption for a general explanation of box plots. locity is a poorer representation of systemic velocities thanthe water maser peak velocity.Caswell & Phillips (2008) and Caswell & Breen (2010)suggested that water masers dominated by blue-shiftedemission might represent a short-lived evolutionary phasein the star formation process. To investigate this we havelooked at the number of water maser sources with peakvelocities offset by more than 10 km s − from the associ-ated 6.7-GHz methanol maser central velocity in the asso-ciation categories of with and without accompanying 12.2-GHz emission (they all have to have 6.7-GHz emission so wehave a measure of the systemic velocity) indicating less andmore evolved sources, respectively (e.g Breen et al. 2010a).We find that in the case of the sources with associated 12.2-GHz sources, there are six sources blue-shifted by more than10 km s − and four that are red-shifted by 10 km s − , com-pared with nine and two in the case of sources with no ac-companying 12.2-GHz emission, lending support to the sug-gestion of Caswell & Phillips (2008) and Caswell & Breen(2010) that water masers showing dominant blue-shiftedemission are indicative of a particularly young high-massstar formation region (although we note we are only dealingwith a small number of sources).In general, the average of the LHCP and RHCP peak ve-locities of excited-state OH masers are in much closer agree-ment with the central velocity of 6.7-GHz methanol masers.Of the 43 excited state OH masers with associated 6.7-GHzmasers, 86 per cent of the sources have peak velocities within ± − (Avison et al. 2016, found similar results usingtheir full sample) and only one source, G 336.983 − − (at 17.1 km s − ). Itis possible that this indicates that the 6.7-GHz methanolmaser emission is only aligned with the excited-state OH maser by chance even though their measured angular sepa-rations is only 0.4 (cid:48)(cid:48) . Among the other five sources with veloc-ity separations of 5 km s − or more, only two have angularseparations greater than 0.4 (cid:48)(cid:48) , at 0.9 and 1.9 (cid:48)(cid:48) , respectively.All six of the excited-state OH sources that have peak ve-locities separated by more than ± − from the central6.7-GHz velocity are redshifted. Within the 100 square degree region of the Galactic plane,the only two masers with peak flux densities greater than1000 Jy are the 6.7-GHz methanol masers G9.621+0.196and G323.740 − ∼ ∼ − − ∼ σ noise levels of each of the searches. Inthe case of the 6.7-GHz methanol masers, the possibility ofa real flux density turn over will be addressed by the moresensitive ‘piggyback’ survey (Ellingsen et al. in prep) whichwas described in Green et al. (2009). Given the sensitivitylimitations it is difficult to confidently comment on any dif-ferences in the peak flux density distributions beyond the ob-vious ranking in peak flux densities from 6.7-GHz methanol,water, 12.2-GHz methanol and, finally, excited-state OH.The lower panel of Fig. 6 shows the distributions of in-tegrated flux densities for the 6.7- and 12.2-GHz methanolmasers (these values are not available for either the watermasers or the excited-state OH masers) and the overall dis-tributions are similar to that of the peak flux densities. Themost striking difference in the distributions is the prepon-derance of 12.2-GHz methanol masers with low integratedflux densities, dominated by a large number of weak, sin- c (cid:13) , 000–000 MB versus HOPS Peak flux density (Jy) N u m be r c oun t s Integrated flux density (Jy km/s) N u m be r c oun t s Figure 6. (Top) Peak flux density distribution of the 6.7-GHzmethanol (magenta; Caswell et al. 2010; Green et al. 2010; Caswellet al. 2011; Green et al. 2012; Breen et al. 2015), 12.2-GHzmethanol (pink; Breen et al. 2012a,b, 2014, 2016), water (blue;Walsh et al. 2014) and excited-state OH masers (orange; Avisonet al. 2016), and; (bottom) integrated flux density distribution ofthe 6.7-GHz methanol (magenta; Breen et al. 2015) and 12.2-GHzmethanol (pink; Breen et al. 2012a,b, 2014, 2016). gle feature masers which is not replicated in the 6.7-GHzdistribution.Fig. 7 shows box plots of the peak flux density distribu-tion of each of the four maser types in association categoriesof solitary (those masers not associated with any of the threeother maser types), associated (those masers associated withone or more of the other maser types) and then associationswith each of the maser types separately. In every case thesolitary masers show the lowest peak flux densities and simi-larly for the 12.2-GHz sources (which are never solitary) thelowest peak flux density distribution are for those sourcesnot associated with either excited-state OH masers or watermaser emission. In each case the highest median and 75thpercentile value is for those masers that are associated withexcited-state OH emission and the second highest median value is for those sources that are associated with 12.2-GHzmethanol maser emission. Fig. 8 shows the luminosity of 6.7-GHz methanol masersin a number of different association categories. The cate-gories of 6.7-GHz methanol masers with no accompanying12.2-GHz emission and with accompanying water but no12.2-GHz masers show the lowest luminosities. The distri-butions of 6.7-GHz methanol maser luminosity extends tomuch higher values for those sources that are associated with12.2-GHz methanol masers and even higher still for sourceswith both 12.2-GHz and water maser emission. 6.7-GHzmethanol maser emission with accompanying excited-stateOH emission shows a distribution similar to those sourceswith 12.2-GHz.Breen et al. (2011) found evidence that 6.7-GHzmethanol masers increase in luminosity as they evolve, sug-gesting that Fig. 8 might imply the following evolution-ary scenario: 6.7-GHz methanol masers devoid of 12.2-GHzmethanol maser emission are the youngest and 6.7-GHzmethanol masers showing 12.2-GHz methanol or excited-state OH masers are present at a slightly later phase of evo-lution. The presence of a water maser may signal a slightlylater evolutionary phase. The population distributions of each maser transition areplotted in a series of histograms in Fig. 9 and 10 as a func-tion of Galactic longitude and latitude, respectively. Thedistributions are similar in all cases, and indeed K-S testsshow that there is no statistically significant evidence thatany of the maser populations are drawn from different distri-butions. However, we note that the G 335 to G 340 longituderange shows the greatest abundance of each of the four masertypes and corresponds to the Perseus arm origin. Qualita-tively it also appears that the two methanol maser distri-butions are very similar to each other, but slightly differentfrom the water and excited-state OH distributions which arealso much more similar to each other. Perhaps the most pro-nounced difference in the distributions is the relative over-abundance of both excited-state OH and water masers inthe latitude range − ◦ to − ◦ compared to the 6.7-and 12.2-GHz methanol masers. A similar feature is seen inthe longitude distributions between longitudes G 305 ◦ andG 310 ◦ , approximately corresponding to the Crux-Scutumarm tangent, but does not correspond to the same sourcesin the latitude range − ◦ to − ◦ .Fig. 10 shows some evidence that the 6.7- and 12.2-GHzmethanol masers are much more tightly constrained to lowerlatitudes than either the water masers or the excited-stateOH masers. Although broader latitude coverage would beneeded to confirm this, it is consistent with the idea that asignificant fraction of the methanol masers are tracing a gen-erally earlier phase of star formation than the other masertypes. In the case of the water masers, a broader latitudedistribution might be further contributed to by their asso-ciation with a broader mass range. c (cid:13) , 000–000 S. L. Breen et al. Peak flux density (Jy) w a t e r e x O H ss o cs o l Peak flux density (Jy) w a t e r e x O H a ss o c . Peak flux density (Jy) w a t e r . ss o cs o l Excited−state OH Peak flux density (Jy) e x O H . ss o cs o l water Figure 7. Peak flux density box plots of 6.7-GHz methanol masers (top left), 12.2GHz methanol masers (top right), excited-state OHmasers (bottom left) and water masers (bottom right) in the categories of solitary (‘sol’; grey), associated with one or more of the othertransitions (‘assoc’; cyan), associated with 6.7-GHz methanol (‘6.7’; magenta), 12.2-GHz (‘12’; pink), excited-state OH (‘exOH’; orange)and water masers (‘water’; blue). Note that in the case of 12.2-GHz methanol masers ‘assoc’ includes those 12.2-GHz sources that areassociated with either or both of excited-state OH or water maser emission since all are associated with 6.7-GHz methanol maser sources.All plots have been truncated in order to see details: the 6.7-GHz plot is missing nine sources, the 12.2-GHz plot is missing six sources,the excited-state OH plot is missing one source and the water plot is missing six sources. See Fig. 3 caption for a general explanation ofbox plots. Green et al. (2011) identified several regions of 6.7-GHzmethanol maser density enhancements, corresponding to sig-nifiant structures within the Galaxy (such as arm origins).Within the HOPS region, there are 224 6.7-GHz methanolmasers that are likely to be associated with these struc-tures (see table 1 of Green et al. 2011, for longitude, lat-itude and velocity ranges corresponding to each region ofenhancement). We find that the 6.7-GHz methanol sourcesthat are associated with these structures have similar as-sociation rates with 12.2-GHz, excited-state OH and watermasers to the entire region (47.8, 6.3, 27.2 per cent associ-ation rates, respectively compared to 45.9, 7.1 and 29.8 percent for the 6.7-GHz methanol masers that are not asso-ciated with major structures). We have also compared theassociation rates of 12.2-GHz, excited-state OH and watermaser emission associated with the 45 6.7-GHz methanolmasers from the MMB survey found to be associated withthe 3-kpc arms (Green et al. 2010). We find that the as-sociation rates with 12.2-GHz, excited-state OH and watermasers are all lower, but given the small numbers, essentiallyequivalent to the larger survey region.Fig. 10 shows that the latitude distribution of the differ-ent maser species all peak at negative values. This is a result that has been found in an increasing large number of Galac-tic plane surveys; both the Bolocam Galactic Plane Survey(BGPS; Rosolowsky et al. 2010) and the APEX TelescopeLarge Area Survey of the GALaxy (ATLASGAL; Schulleret al. 2009; Beuther et al. 2012; Contreras et al. 2013) sur-veys of dust continuum emission found a latitude distribu-tion peak at − ◦ . Reed (2006) measured the location ofthe Sun to be ∼ 20 pc above the Galactic plane, which willskew sources within the plane of the Galaxy to negativevalues. Indeed, both Schuller et al. (2009) and Contreraset al. (2013) attribute the excursion of their sources from theGalactic plane to the location of the Sun above the plane.The 6.7-GHz methanol masers have a mean latitude of − ◦ ± ◦ , the excited-state OH masers have a meanlatitude of − ◦ ± ◦ and the water masers have a meanlatitude of − ◦ ± ◦ . The offset between the dust con-tinuum surveys and the 6.7-GHz methanol masers is likelybecause the MMB is much more sensitive to sources onthe far side of the Galaxy which do not show the sameoffset from the Galactic plane. In fact, the latitude dis-tribution to 6.7-GHz sources located at distances fartherthan 8 kpc shows a mean latitude that is consistent withzero ( − ◦ ± ◦ ). Splitting these distant sources into c (cid:13)000 20 pc above the Galactic plane, which willskew sources within the plane of the Galaxy to negativevalues. Indeed, both Schuller et al. (2009) and Contreraset al. (2013) attribute the excursion of their sources from theGalactic plane to the location of the Sun above the plane.The 6.7-GHz methanol masers have a mean latitude of − ◦ ± ◦ , the excited-state OH masers have a meanlatitude of − ◦ ± ◦ and the water masers have a meanlatitude of − ◦ ± ◦ . The offset between the dust con-tinuum surveys and the 6.7-GHz methanol masers is likelybecause the MMB is much more sensitive to sources onthe far side of the Galaxy which do not show the sameoffset from the Galactic plane. In fact, the latitude dis-tribution to 6.7-GHz sources located at distances fartherthan 8 kpc shows a mean latitude that is consistent withzero ( − ◦ ± ◦ ). Splitting these distant sources into c (cid:13)000 , 000–000 MB versus HOPS Luminosity (Jy kpc^2) e x O Hw a t e r + w a t e r − + − Figure 8. Peak luminosity of 6.7-GHz methanol masers with(+12; pink) and without (-12; magenta) accompanying 12.2-GHzmethanol masers emission, with water but no 12.2-GHz (water-12; purple), with both 12-GHz and water (water+12; violet) andwith excited-state OH maser emission (exOH; orange). Luminosi-ties have been calculated using distances from Green & McClure-Griffiths (2011), Green et al. (2017) and references therein (whereavailable). There are 16 6.7-GHz methanol masers with luminosi-ties greater than 10000 Jy km s − and are therefore excludedfrom the plot (470 have luminosities less than 10000). See Fig. 3caption for a general explanation of box plots. two groups based on quadrants, results in mean latitudevalues of − ◦ ± ◦ for the first quadrant sources and0.004 ◦ ± ◦ for the fourth quadrant, consistent with theexpectations of the Galactic warp. Alongside the water maser observations, HOPS simultane-ously observed a number of ammonia transitions (Walshet al. 2011). Purcell et al. (2012) presented a catalogue ofthe ammonia (1,1) and (2,2) from the HOPS survey, whichhas a mean sensitivity of 0.20 K and catalogues 669 densemolecular clouds. Using these data, Longmore et al. (2017)investigated the properties of the dense gas and made somecomparisons between those sources with and without 6.7-GHz methanol masers and HOPS water masers. Longmoreet al. (2017) were able to derive kinetic temperatures for64 ammonia sources and they looked at maser associationsand found that those sources with either 6.7-GHz methanolmasers or water masers were warmer than those without andalso had larger linewidths. P e r c en t age o f popu l a t i on 30 20 10 0 350 340 330 320 310 300 290 P e r c en t age o f popu l a t i on 30 20 10 0 350 340 330 320 310 300 290 P e r c en t age o f popu l a t i on 30 20 10 0 350 340 330 320 310 300 290 excited−state OH Galactic longitude (degrees) P e r c en t age o f popu l a t i on 30 20 10 0 350 340 330 320 310 300 290 water Figure 9. Histograms of the percentage of 6.7-GHz methanol,12.2-GHz methanol, excited-state OH and water masers as a func-tion of Galactic longitude.c (cid:13) , 000–000 S. L. Breen et al. P e r c en t age o f popu l a t i on −0.5 −0.3 −0.1 0 0.1 0.2 0.3 0.4 0.5 P e r c en t age o f popu l a t i on −0.5 −0.3 −0.1 0 0.1 0.2 0.3 0.4 0.5 P e r c en t age o f popu l a t i on −0.5 −0.3 −0.1 0 0.1 0.2 0.3 0.4 0.5 excited−state OH Galactic latitude (degrees) P e r c en t age o f popu l a t i on −0.5 −0.3 −0.1 0 0.1 0.2 0.3 0.4 0.5 water Figure 10. Histograms of the percentage of 6.7-GHz methanol,12.2-GHz methanol, excited-state OH and water masers as a func-tion of Galactic latitude. Note that there are only 433 of the starformation water masers plotted here since two of them fall slightlyoutside the latitude range. We have compared the ammonia sources associatedwith any combination of 6.7-, 12.2-GHz methanol, excited-state OH or water maser emission and find that those associ-ated with masers have a higher average and median tempera-ture (28.5 K and 24 K) compared to those without any maseremission (25.5 K and 22 K), although due the significantoverlap in the temperatures and the small sample size, thisdifference is not statistically significant. Since there are only64 sources in the Longmore et al. (2017) sample, we foundthat splitting the sources into association categories basedon the different combinations of detected masers yielded nomeaningful results. Guzm´an et al. (2015) calculated far-infrared dust temper-atures using Hi-Gal (Herschel Infrared Galactic Plane Sur-vey) and ATLASGAL (APEX Telescope Large Area Sur-vey of the Galaxy) data, covering wavelengths between 160-and 870- µ m. While their work investigated the dust tem-peratures of ∼ µ m maps to ∼ (cid:48)(cid:48) .Fig. 11 shows the far-infrared temperature distributionof regions associated with each of the four types of masers.Although there is a lot of overlap in each of the groups -both because some sources are present in more than oneof the categories and also because temperatures have beenaveraged across significant regions of sky - there are statis-tically significant differences between the groups. Accordingto a Wilcoxon rank-sum test there are statistically signifi-cant differences between the far-infrared dust temperaturesat the locations of the excited-state OH masers and eachof the other masers types (p-values all smaller than 0.002)and a similar result is found by running a K-S test, whichshows that the distributions are significantly different (p-values all better than 0.01). None of the other maser cate-gories show temperatures that are statistically significantlydifferent from any of the other groups but this is largely dueto some limitations caused by overlap in the datasets (es-pecially since all of the 12.2-GHz sources are also presentin the 6.7-GHz source category, for example). If we com-pare the temperatures of regions associated with just 6.7-GHz methanol masers to the regions that have both 6.7-and 12.2-GHz emission we find that the temperatures at thelocations of 12.2-GHz masers are significantly warmer (seeTable 3) than those without (Wilcoxon rank-sum gives a p-value of 0.05 and a K-S test shows that the distributions aresignificantly different, with a p-value of 0.02).Class II methanol masers have provided some of theclearest maser evolutionary trends to date, and is thoughtto be a result of the close connection between the masersand the protostar ensured by radiative pumping. In thisevolutionary scenario, 6.7-GHz methanol masers appear sig-nificantly earlier than their 12.2-GHz counterparts (Breenet al. 2010b) and, other, rarer class II methanol masers at37.7-GHz methanol masers trace a short-lived phase, presentjust prior to the cessation of the class II methanol maserphase (Ellingsen et al. 2011, 2013). That the far-infrared c (cid:13) , 000–000 MB versus HOPS Temperature (K) w a t e r e x O H . 20 30 40 50 Figure 11. Far infrared dust temperatures of regions exhibit-ing 6.7-GHz methanol masers (6.7; magenta), 12.2-GHz methanolmasers (12; pink), excited-state OH masers (exOH; orange) andwater masers (water; blue). See Fig. 3 caption for a general ex-planation of box plots. Table 3. Mean and median dust temperatures in a number of as-sociation categories; in the first section 6.7-GHz methanol maserswith and without associated 12.2-GHz methanol maser emissionare shown. The standard deviation is given in column four. Thesecond and third vertically separated sections correspond to thedata shown in Fig 12.Association Mean Median StandardCategory Temp (K) Temp (K) deviation6.7-GHz with 12.2-GHz 24.7 24.6 4.26.7-GHz no 12.2-GHz 24.2 23.8 4.8solitary 6.7-GHz 23.6 23.0 4.2associated 6.7-GHz 24.9 24.8 4.66.7-GHz with 12.2-GHz 24.7 24.6 4.26.7-GHz with water 25.4 25.2 4.86.7-GHz with exOH 26.7 27.0 5.0solitary water 23.8 23.0 5.8associated water 25.6 25.2 5.0water with 6.7-GHz 25.4 25.2 4.8water with 12-GHz 25.5 25.8 3.8water with exOH 26.9 26.7 6.3 dust temperatures are higher for 6.7-GHz masers associ-ated with 12.2-GHz masers than 6.7-GHz masers devoid ofaccompanying 12.2-GHz emission, provides an independentcorroboration that 12.2-GHz methanol masers are presentat a later evolutionary phase. It is, of course, possible thatmass is a confounding variable here, but given the resolutionof 35 (cid:48)(cid:48) (corresponding to a linear resolution of 0.85 pc at adistance of 5 kpc), evolution is a much more likely culpritgiven that it will take a significant time to heat the sur-rounding medium enough to result in a spatially averagedtemperature difference.Fig. 12 shows the temperatures of sources exhibiting6.7-GHz methanol and water masers, respectively, in eachof the association categories. The distribution of solitarysources are skewed towards lower temperatures in both cases, but interestingly the temperature of the solitary 6.7-GHz masers are not drawn from the same population as thesolitary water masers (p-value from K-S test 0.03). Table 3shows that there is little difference in the median and meanvalues of these categories but Fig. 12 shows that the tem-peratures of regions associated with solitary water maserscover a broader range than those associated with solitary6.7-GHz masers (interquartile ranges of 8.0 K and 5.3 K,respectively).In each case, the association categories show statisti-cally significant differences in temperature compared to soli-tary water or methanol masers (resultant p-values from K-Stests are all smaller than 0.002, except for water masersthat also show excited-state OH maser emission which hasa p-value of 0.048). Additionally we find that the associa-tion categories of 6.7-GHz with accompanying 12.2-GHz aredrawn from a different population than the 6.7-GHz withexcited-state OH masers (p-value 0.015).Breen et al. (2011) found that both the luminosity andthe velocity ranges of sources increased with evolution, sug-gesting in the latter case that this was a result of eitherincreased internal motions or that the volume of gas con-ducive to the maser action increased as the sources evolved.We have compared the dust temperatures of the 6.7-GHzmethanol maser sources with the top 20 per cent of velocityranges to the rest of the sample and find that the sourceswith the highest velocity ranges have both higher average(24.8 K compared with 24.3 K) and median (24.8 K com-pared with 24.0 K) temperatures.We find a similar result if we look at 6.7-GHz methanolmaser luminosity instead (which is expected given the rela-tionship found between maser velocity range and luminosity)- the most luminous 20 per cent of maser sources (that havetemperature data) are warmer, on average (25.1 comparedwith 24.1 K with medians of 25.1 and 23.8 K). While theseresults are all self consistent, none of the comparisons madebetween velocity ranges or luminosities and dust temper-atures are statically significant, meaning that average andmedian values are higher, but there is no statistical evidencethere is a population difference.Since the resolution of the temperature and column den-sity maps is relatively low (35 arcsec) compared to individ-ual young high-mass stars, the small temperature differenceis most likely representing a much larger temperature dif-ference that has been spatially averaged, diluting the truedifferences. The fact that there is no tight one-to-one corre-lation in any of these of these comparisons is likely a conse-quence of this. Comparisons between expectations from maser modellingand measured physical conditions with respect to maser oc-currence has the potential to significantly increase our un-derstanding of the masers themselves, as well as the regionsin which they are located. That we find statistically dif-ferent dust temperatures between sources showing excited-state OH masers compared to the other maser types, as wellas 6.7-GHz methanol masers with and without accompany-ing 12.2-GHz methanol masers, gives us a unique opportu-nity to compare with maser models (e.g. Cragg, Sobolev &Godfrey 2002, 2005). c (cid:13) , 000–000 S. L. Breen et al. Temperature (K) w a t e r e x O H ss o cs o l 15 20 25 30 35 40 45 50 Temperature (K) e x O H . ss o cs o l 20 30 40 50 water Figure 12. Temperatures of the regions showing (top) just 6.7-GHz methanol maser emission (‘sol’; grey), as well as those thatare associated with one or more maser type (‘assoc’; cyan) andassociations with 12.2-GHz methanol masers (‘12’; pink), excited-state OH masers (‘exOH’; orange) and water masers (‘water’;blue) and (bottom) just water maser emission (‘sol’; grey), as wellas those that are associated with one or more maser type (‘assoc’;cyan) and associations with 6.7-GHz methanol masers (‘6.7’; ma-genta), 12.2-GHz methanol masers (‘12’; pink) and excited-stateOH masers (‘exOH’; orange). See Fig. 3 caption for a generalexplanation of box plots. Cragg, Sobolev & Godfrey (2005) presented maser mod-elling of class II methanol masers and their figs 2 and 3 showhow maser brightness varies with a number of parameters,including dust temperature, which shows a sharp increase inmaser brightness at a higher dust temperature for 12.2-GHzmethanol masers compared to 6.7-GHz methanol masers.Their findings are naively consistent with our results, show-ing that sources associated with both 6.7- and 12.2-GHzmethanol masers have slightly warmer dust temperatures,however, there are a number of complex factors at play, in-cluding the fact that these parameter plots show how dusttemperature varies as other parameters remain fixed, andthe fact that our dust temperatures represent the average ofa significant region.Cragg, Sobolev & Godfrey (2002) considered the char-acteristics of methanol and OH maser emission in their mod- els, including the 6.7 and 12.2-GHz transitions of methanoland the 6035-MHz transition of OH. They found that6.7-GHz methanol masers appear at dust temperatures of ∼ 100 K and then decline as the gas temperature increasesto that of the dust temperature, while the 6035-MHz excitedstate OH masers was quenched at gas temperatures that sur-passed 70 K but were independent of dust temperature. Itis somewhat difficult to reconcile these predictions with ourresults, given that they are largely based on gas tempera-ture, which our low resolution dust temperatures can shedlittle light on. We have compared the occurrence of 6.7-GHz and 12.2-GHzmethanol masers with 6035-MHz excited-state OH and 22-GHz water masers within the 100 square degree region com-mon to HOPS and the MMB survey. We find that thereare 634 6.7-GHz methanol masers, 435 water masers, 29512.2-GHz methanol masers and 80 excited state OH maserswithin the survey range. The water maser population is ef-fected by the lower sensitivity of HOPS and we expect thata water maser survey that had a similar sensitivity to theMMB were conducted, the number would be significantlyincreased, perhaps to a level that exceeded the 6.7-GHzmethanol maser population.Water masers show the highest median velocityrange (9.8 km s − ), followed by 6.7-GHz methanol masers(6.1 km s − ), excited-state OH (4.3 km s − ) and 12.2-GHzmethanol masers (1.9 km s − ). We additionally find that themedian velocity range of each of the maser types is lowerwhen considering only sources that are not associated withother maser types. More than one third of the solitary watermasers exhibit only one spectral feature.The luminosity of 6.7-GHz methanol maser emissionis lowest when these masers are either solitary, or onlyassociated with water maser emission. 6.7-GHz methanolmasers associated with 12.2-GHz emission or excited-stateOH maser emission are much more luminous (whether ornot they are also accompanied by water maser emission).We find that 89 per cent of water masers exhibit theirpeak emission within ± 10 km s − of the 6.7-GHz methanolmaser central velocity (which gives a reliable indication ofthe systemic velocity) and that water maser peak velocity is,in general, close to the systemic velocity than their centralvelocity. Excited-state OH masers generally have velocitiesmuch closer to the systemic velocity - 86 per cent within ± − of the 6.7-GHz methanol maser central velocity.All six of the excited-state OH masers with greater velocityseparations show redshifted velocities.Comparison of the far-infrared dust temperatures ofsources exhibiting each of the maser types shows that thoseexhibiting excited-state OH masers are statistically signif-icantly warmer. Since sources can exhibit more than oneof the four types of masers (resulting in some sources be-ing present in each of the associated categories) we alsolooked explicitly at a number of additional categories. Ineach case, sources exhibiting solitary water or solitary 6.7-GHz methanol masers have lower far-infrared dust temper-atures than sources exhibiting additional types of maseremission. We also found that the far-infrared dust temper- c (cid:13) , 000–000 MB versus HOPS atures associated with sources exhibiting both 6.7-GHz and12.2-GHz sources were significantly warmer than those onlyshowing 6.7-GHz maser emission. These results provide inde-pendent support to the idea that different masers are presentat different stages in the evolution of high-mass star forma-tion regions. ACKNOWLEDGMENTS We thank A. E. Guzm´an for providing us with far-infrareddust temperature maps of the Galactic plane. The Parkestelescope, the Australia Telescope Compact Array and Mo-pra telescope are part of the Australia Telescope which isfunded by the Commonwealth of Australia for operation as aNational Facility managed by CSIRO. Financial support forthis work was provided by the Australian Research Council.This research has made use of: NASA’s Astrophysics DataSystem Abstract Service; and the SIMBAD data base, op-erated at CDS, Strasbourg, France. REFERENCES Avison, A., et al., 2016, MNRAS, 461, 136Beuther H., Walsh A., Schilke P., Sridharan T. K., MentenK. M., Wyrowski F., 2002, A&A, 390, 289Beuther H., et al., 2012, ApJ, 747, 43B(cid:32)laszkiewicz L., Kus A. J., 2004, A&A, 413, 233Brand J., Cesaroni R., Comoretto G., Felli M., Palagi F.,Palla F., Valdettaro R., 2003, A&A 407, 573Breen S. L., Ellingsen S. P., Caswell J. L., Lewis B. J.,2010, MNRAS, 401, 2219Breen S. L., Caswell J. L., Ellingsen S. P., Phillips C. J.,2010b, MNRAS, 406, 1487Breen S. L., Ellingsen S. P., 2011, MNRAS, 416, 178Breen S. L., Ellingsen S. P., Caswell J. L., Green J. A.,Voronkov M. A., Fuller G. A., Quinn L. J., Avison A.,2011, ApJ, 733, 80Breen S. L., Ellingsen S. P., Caswell J. L., Green J. A.,Voronkov M. A., Fuller G. A., Quinn L. J., Avison A.,2012a, MNRAS, 421, 1703Breen S. L., Ellingsen S. P., Caswell J. L., Green J. A.,Voronkov M. A., Fuller G. A., Quinn L. J., Avison A.,2012b, MNRAS, 426, 2189Breen, S. L., Ellingsen S. P., Contreras Y., Green J. A.,Caswell J. L., Stevens J. B., Dawson J. R., Voronkov M.A., 2013, MNRAS, 435, 524Breen S. L., Ellingsen S. P., Caswell J. L., Green J. A.,Avison A., Voronkov M. A., Fuller G. A., Quinn L. J.,Titmarsh, A., 2014, MNRAS, 438, 3368Breen S. L., et al., 2015, MNRAS, 450, 4109Breen S. L., et al., 2016, MNRAS, 459, 4066Caswell J. L., Gardner F. F., Norris R. P., Wellington, K.J., McCutcheon W. H., Peng, R. S., 1993, MNRAS, 260,425Caswell J. L., Vaile R. A., Ellingsen S. P., Norris R. P.,1995, MNRAS, 274, 1126Caswell J. L., 1997, MNRAS, 289, 203Caswell J.L., Phillips C.J., 2008, MNRAS, 386, 1521Caswell J. L., 2009, PASA, 26, 454Caswell J. L. et al., 2010, MNRAS, 404, 1029 Caswell J. L. et al., 2011, MNRAS, 417, 1964Caswell J. L., Breen S. L., 2010, MNRAS, 407, 2599Claussen M. J., Wilking B. A., Benson P. J., Wootten A.,Myers P. C., Terebey S., 1996, ApJS, 106, 111Contreras Y., et al., 2013, A&A, 549, 45Cragg D. M., Sobolev A. M., Godfrey P. D., 2002, MNRAS,331, 521Cragg D. M., Sobolev A. M., Godfrey P. D., 2005, MNRAS,360, 533Dame T. M., Hartmann D., Thaddeus P., 2001, ApJ, 547,792Deacon R. M., Chapman J. M., Green A. J., Sevenster M.N., 2007, ApJ, 658, 1096Ellingsen S. P., Voronkov M. A., Cragg D. M., Sobolev A.M., Breen S. L., Godfrey P. D., 2007, in Chapman J. M.,Baan W. A., eds., Proc. IAU Symp., 242, AstrophysicalMasers and their Environments. Cambridge Univ. Press,Cambridge, p. 213Ellingsen S. P., Breen S. L., Sobolev A. M., Voronkov M.A., Caswell J. L., Lo N., 2011, ApJ, 742, 109Ellingsen S. P., Breen S. L., Voronkov M. A., Dawson J.R., 2013, MNRAS, 429, 3501Ellingsen S. P., et al., (in prep) Felli M., Brand J., Cesaroni R., Codella C., Comoretto G.,di Franco S., Massi F., Moscadelli L., Nesti R., Olmi L.,Palagi F., Panella D., Valdettaro R., 2007, A&A, 476, 373Furuya R. S., Kitamura Y., Wootten H. A., Claussen M.J., Kawabe R., 2001, ApJL, 559, 143Gaylard M. J., MacLeod G. C., van der Walt D. J., 1994,MNRAS, 269, 257Green J. A. et al., 2009, MNRAS, 392, 783Green J. A. et al., 2010, MNRAS, 409, 913Green J. A. et al., 2011, ApJ, 733, 27Green J. A., McClure-Griffiths, N., 2011, MNRAS, 417,2500Green J. A. et al., 2012, MNRAS, 420, 3108Green J. A. et al., 2017, MNRAS, 469, 1383Guzm´an A. E., Sanhueza P., Contreras Y., Smith H. A.,Jackson J. M., Hoq S., Rathborne J. M., 2015, ApJ, 815,130Jackson J. M., et al. 2013, PASA, 30, 57Kemball A. J., Gaylard M. J., Nicolson G. D., 1988, ApJ,331, L37Koo B. C., Williams D. R. D., Heiles C., Backer D. C, 1988,ApJ, 326, 931Longmore S. N., et al. 2017, MNRAS in press Minier V., Ellingsen S. P., Norris R. P., Booth R. S., 2003,A&A, 403, 1095Moscadelli L. et al., 2002, ApJ., 564, 813Norris R. P., Caswell J. L., Gardner F. F., Wellington K.J., 1987, ApJ, 321, L159Norris R. P., Whiteoak J. B., Caswell J. L., Wieringa M.H., Gough R. G., 1993. ApJ, 412, 222Pandian J. D., Menten K. M., Goldsmith P. F., 2009, ApJ,706, 1609Pandian J. D., Goldsmith P. F., Deshpande A. A., 2007,ApJ, 656, 255Purcell C. R., et al., 2012, MNRAS, 426, 1972Reed B. C,. 2006, J. R. Astron. Soc. Can., 100, 146Reid M. J., Schneps M. H. Moran J. M. Gwinn C. R. GenzelR., Downes D., Roennaeng B., 1988, ApJ, 330, 809Rosolowsky E. et al., 2010, ApJS, 188, 123 c (cid:13) , 000–000 S. L. Breen et al. Schuller F., et al., 2009, A&A, 504, 415Szymczak M., Pillai T., Menten K. M., 2005, A&A, 434,613Szymczak M., Bartkiewicz, A., Richards, A. M. S., 2007,A&A, 468, 617Titmarsh A. M., Ellingsen S. P., Breen S. L., Caswell J. L.,Voronkov M. A., 2014, MNRAS, 443, 2923Titmarsh A. M., Ellingsen S. P., Breen S. L., Caswell J. L.,Voronkov M. A., 2016, MNRAS, 459, 157Walsh A. J., et al. 2014, MNRAS, 442, 2240Walsh A. J., et al. 2011, MNRAS, 416, 1764Wilson W. E. et al., 2011, MNRAS, 416, 832Xu Y., Li J. J., Hachisuka K., Pandian J. D., Menten K.M., Henkel C., 2008, A&A, 485, 729 c (cid:13)000