Pre-main sequence accretion in the low metallicity Galactic star-forming region Sh 2-284
DDraft version December 5, 2014
Preprint typeset using L A TEX style emulateapj v. 05/12/14
PRE-MAIN SEQUENCE ACCRETION IN THE LOW METALLICITYGALACTIC STAR-FORMING REGION Sh 2-284
V. M. Kalari & J. S. Vink
Armagh Observatory, College Hill, Armagh, BT61 9DG, UK; [email protected]
Draft version December 5, 2014
ABSTRACTWe present optical spectra of pre-main sequence (PMS) candidates around the H α region takenwith the Southern African Large Telescope, SALT, in the low metallicity ( Z ) Galactic region Sh2-284, which includes the open cluster Dolidze 25 with an atypical low metallicity of Z ∼ Z (cid:12) .It has been suggested on the basis of both theory and observations that PMS mass-accretion rates,˙ M acc , are a function of Z . We present the first sample of spectroscopic estimates of mass-accretionrates for PMS stars in any low- Z star-forming region. Our data-set was enlarged with literaturedata of H α emission in intermediate-resolution R -band spectroscopy. Our total sample includes 24objects spanning a mass range between 1 - 2 M (cid:12) and with a median age of approximately 3.5 Myr.The vast majority (21 out of 24) show evidence for a circumstellar disk on the basis of 2MASS and Spitzer infrared photometry. We find ˙ M acc in the 1 - 2 M (cid:12) interval to depend quasi-quadratically onstellar mass, with ˙ M acc ∝ M . ± . ∗ , and inversely with stellar age ˙ M acc ∝ t − . ± . ∗ . Furthermore,we compare our spectroscopic ˙ M acc measurements with solar Z Galactic PMS stars in the same massrange, but, surprisingly find no evidence for a systematic change in ˙ M acc with Z . We show thatliterature accretion-rate studies are influenced by detection limits, and we suggest that ˙ M acc may becontrolled by factors other than Z ∗ , M ∗ , and age. Subject headings: stars: pre main-sequence, stars:variables: T Tauri, Herbig Ae/Be, accretion disks,stars: formation, stars: fundamental parameters INTRODUCTION
Present understanding suggests that most stars accretemass from a circumstellar disk during their pre-main se-quence (PMS) phase (see Hartmann 2008). The rateof mass accretion ( ˙ M acc ) from the disk to the star isvital to describe the system’s evolution, including thepotential growth of planets in the disk whilst the starreaches its main sequence configuration. Viscous diskevolution models predict that ˙ M acc drops with time as˙ M acc ∝ t ∗− . , a prediction which has been substantiatedby empirical studies (Hartmann et al. 1998). Observa-tions have also found that the mass-accretion rate scaleswith stellar mass as ˙ M acc ∝ M ∗ α over a mass range from0.3 M (cid:12) -3 M (cid:12) , with index α ∼ M ∗ < M (cid:12) (Fang et al. 2009; Barentsenet al. 2011). Note that the ˙ M acc vs. Mass relation is notreproduced by the current paradigm of disk evolution,and it is subject to ongoing research (e.g. Hartmann etal. 2006).It has also been suggested that ˙ M acc is a function ofmetallicity, Z . Seven PMS star candidates discoveredin the Large Magellanic Cloud (LMC) at Z ≈ Z region. Beaulieu et al. (1996) se-lected these objects because they exhibited peculiar pho-tometric variability similar to Galactic Herbig HAe/Bestars. The number of low- Z PMS candidates was in-creased in the follow-up works of Lamers et al. (1999); deWit et al. (2002); de Wit et al. (2003) and de Wit et al. (2005) using similar techniques, where the authors notedthat these stars also exhibited H α emission and were lo-cated in H II regions. These studies showed that the lu-minosities of these objects were significantly higher thanthose of Galactic PMS stars at solar Z (cid:12) , and with similarspectral types. de Wit et al. (2003) showed that HAeBecandidates in the Small Magellanic cloud (SMC) at lowermetallicities ( Z ∼ M acc are analogously related to metallicity is not clear.The ˙ M acc estimated by de Wit et al. (2005) for theirbest LMC PMS candidate (ELHC 7) indicated no signif-icant difference in comparison to the canonical Galacticaverage.On the contrary, studies based on the Balmer contin-uum excess (Romaniello et al. 2004) and H α emission(Panagia et al. 2000) inferred higher ˙ M acc compared tothe Galactic average in a LMC region close to SN1987A.Follow-up H α photometric studies by the same group inthe LMC regions around SN1987A (De Marchi et al.2010) and 30 Doradus (De Marchi et al. 2011a) whichwere summarized in Spezzi et al. (2012); and in theSMC open clusters NGC 346 ( Z ∼ Z ∼ M acc (seebelow) whilst also finding ˙ M acc ∝ M ∗ , and ˙ M acc ∝ t ∗ . in the 1-2 M (cid:12) range for the LMC PMS stars. They em- a r X i v : . [ a s t r o - ph . S R ] D ec V. M. Kalari & J. S. Vinkployed a novel method using VI H α photometry, ratherthan conventional spectroscopy to derive ˙ M acc . We haveapplied a similar method in the Galactic regions IC 1396(Barentsen et al. 2011), NGC 2264 (Barentsen et al.2013) and NGC 6530 (Kalari et al. 2014, in prep.), andfound that the results from the photometric method donot differ significantly from spectroscopic H α or U -bandmeasurements, nor did we find a large number of inter-lopers. Furthermore, we discovered an accreting PMScandidate star in the LMC region 30 Doradus (Kalariet al. 2014), which is within ∼
10 pc to one of the re-gions examined in Spezzi et al. (2012). These vari-ous results suggest there may well be genuine highly-accreting PMS stars present in the MCs. De Marchi etal. (2013) suggest accretion rates are inversely propor-tional to metallicity based on studies of the two SMCopen clusters. De Marchi et al. (2011b) reported amedian ˙ M acc of 4 × − M (cid:12) yr − in NGC 346 for a bi-modal age distribution of 1 and 20 Myr; and De Marchiet al. (2013) found a median of 2.5 × − M (cid:12) yr − fora 2 Myr population in NGC 602 and 4 × − M (cid:12) yr − for a 20 Myr population. De Marchi et al. (2013) sum-marized their work in the MCs using an equation of theform log M acc = a × log t ∗ + b × log M ∗ + c , where a, b and c are constants, and c is inversely proportional to Z. Sim-ilarly, Spezzi et al. (2012) suggest that ˙ M acc is inverselyproportional to Z , and suggest that ˙ M acc in the LMC ishigher than that of Galactic PMS stars having similarmasses, irrespective of age. Spezzi et al. (2012) founda median ˙ M acc ∼ × − M (cid:12) yr − of LMC PMS starshaving an average age of 10 Myr, when accretion in mostlow-mass ( < M (cid:12) ) Galactic PMS stars is thought tohave ceased (Fedele et al. 2010). Approximately, for 1 -2 M (cid:12) the PMS lifetime is between 30 - 5 Myr respectively(Palla & Stahler 1999). This suggests that accretion isa dominant process for a large fraction of the PMS life-time in the LMC, contrary to current observations inthe Galaxy. Prolonged accretion at these rates suggestsaccretion adds significant mass to the central star afterthe ’birthline’ (Stahler 1983; Palla & Stahler 1990; Hart-mann et al. 1997; Tout et al. 1999) – contrary to currentdisk or spherical accretion models. The authors suggestthat the lower disk opacity, and viscosity due to lowering Z leads to a longer viscous timescale, thereby allowingaccretion to take place at such older ages.However, observational studies of disk lifetime at lower Z indicate that metal-poor stars in the extreme outerGalaxy ( Z ≈ Z PMS stars (Yasui et al. 2009). Ac-cretion at the high rates measured in the MCs by Spezziet al.(2012) and De Marchi et al.(2010;2011a;2011b;2013)means that the disks of these PMS stars have under-gone gravitational instability (Pringle 1981; Hartmann2006; Cai et al. 2006), shortening the disk lifetime. Fur-thermore, lowering Z lowers the dust shielding efficiencyagainst UV photoevaporation from OB stars leading topossible disk erosion, an effect which has been demon-strated in the LMC (Spezzi et al. 2012). Internal pho-toevaporation is also thought to reduce disk lifetimes atlow- Z , as the lower metal content increases X-ray opac-ity (Ercolano & Clarke 2010). In all these cases disk life-times in low- Z environments are thought to be reduced, suggesting that it is challenging for accretion to proceedat significant rates at older ages in low Z environments.It also suggests that ˙ M acc may be higher at lower Z , butonly at young ages.It is thus essential to further scrutinize the differencesin ˙ M acc of metal-poor PMS stars in comparison to solar Z PMS stars, as to understand star formation in different Z environments, as well as the processes involved in diskevolution. To this purpose, we employ spectroscopic ob-servations of candidate PMS stars located in the Galacticstar-forming region Sh 2-284 (Sharpless 1959). Sh 2-284is an H II region encompassing the central open clusterDolidze 25 (also OCL-537). The B stars of Dolidze 25have been shown to be metal deficient with respect tosolar values, with all measured elemental abundancesranging from − − − M (cid:12) range with medium-resolution spectroscopy, andwe can compare our measured accretion rates with liter-ature spectroscopic measurements of solar Z PMS stars,as well as results claimed from H α photometry in theLMC by Spezzi et al. (2012) .The paper is organized as follows. Section 2 detailsthe data sample, and measured properties of the PMSstars. Section 3 describes our results and method to cal-culate ˙ M acc . A discussion of the ˙ M acc distribution withrespect to M ∗ , t ∗ and Z is presented in Section 4. Theconclusions of our work are summarized in Section 5. DATA
For our program, we selected PMS candidates on thebasis of infrared 2MASS and
Spitzer photometry. Six
Spitzer identified Class II sources Young Stellar Objects(YSO) from Puga et al. (2009) with near-infrared
JHKs magnitudes from 2MASS (Two Micron all sky survey;Cutri et al. 2003) or optical
UBV magnitudes from Del-gado et al. (2010) were selected. This information is pre-sented in Table 1. The location of the selected stars isoverlaid on an H α gray-scale in Fig. 1. Long-slit spectraof these objects were obtained using the Robert StobieSpectrograph (RSS) on the 10 metre Southern AfricanLarge Telescope ( SALT ) located in Sutherland, SouthAfrica (Burgh et al. 2003). A slit width of 1.25 was usedto obtain good sky subtraction. A red grating covering λλ R ∼ .Bias subtraction, flat-field correction, and cosmic rayremoval were performed using standard IRAF proce- Due to the lack of publicly available data of accretion propertiesof PMS stars in the SMC, we restrict our comparison to the LMCPMS sample of Spezzi et al. (2012). IRAF is distributed by the National Optical Astronomy Ob-servatory, which is operated by the Association of Universities forResearch in Astronomy, Inc., under cooperative agreement withthe National Science Foundation. re-main sequence accretion rates at low metallicity 3
TABLE 1Literature data for observed stars α (J2000) δ (J2000) V B − V U − B J J − H J − Ks D10 a P09 b a Identification no. in Delgado et al. (2010). b Identification no. in Puga et al. (2009).
RA (J2000) +0°06'00.0"12'00.0"18'00.0"24'00.0"30'00.0" D e c ( J ) Dol 25 ◦ Fig. 1.—
Sh 2-284 in H α taken from Parker et al. (2005). Northis up and east is to the left. Stars observed with SALT are markedby circles, and stars with
VIMOS spectra are marked by crosses.The central cluster Dolidze 25 is also marked. At d = 4000 pc, 0.15 ◦ corresponds to ∼
10 pc. dures. Wavelength calibration was performed using Nearc spectra using the
IRAF identify task. Spectra werecorrected for sensitivity using the Sutherland extinctioncurve. The variable pupil design of the
SALT telescopemakes absolute flux calibration impossible. The final ex-tracted 1-dimensional spectra are shown in Fig 2. Theaverage signal to noise is ∼
20 - 40.
Basic analysis of the
SALT data
Equivalent widths (EW) of emission and absorptionlines were measured using the
IRAF task splot . Spec-tral types were derived from the EW of the (i) Mg II II I T eff ) were estimated using the spectraltype- T eff calibration of Kenyon & Hartmann (1995). H α line profiles are displayed in Fig. 3. Five stars show H α in emission. Combined with the fact that these stars allshow near to mid-infrared excess resembling PMS stars,we take these five stars to be bonafide accreting PMSstars.Model colors and bolometric corrections were calcu-lated using Castelli & Kurucz (2004) model spectra at Wavelength (˚A) N o r m a li ze d R e l a t i v e flu x J06451007+0014117J06445837+0014151J06443682+0016186
Fig. 2.—
Normalized medium resolution longslit spectra of PMScandidates in Sh 2-284 observed using the
SALT telescope. Fromtop to bottom we show the spectra of an A9, F0 spectral type PMSstars, and a F6 star with no H α emission but infrared-excess. Forthe complete results see Table 2. . . . . . . N o r m a li ze d I n t e n s i t y J06450681+0013535 . . . . . . J06445577+0013168 . . . . . . J06450208+0019443 − −
200 0 200 4000 . . . . . . N o r m a li ze d I n t e n s i t y J06445837+0014151 − −
200 0 200 400
Velocity (km s − ) . . . . . . J06443682+0016186 − −
200 0 200 400 − . − . − . − . − . . . J06451007+0014117
Fig. 3.—
Normalized H α line profiles of the SALT spectra. Thevelocities of all stars with H α in emission is larger than 270 km s − ,indicating clearly accretion in the line profile. The measured H α EW are given in Table 2.
V. M. Kalari & J. S. Vinkthe appropriate T eff and Z . We assumed the logarithmof surface gravity log g = 4.0 ± BV photometry from Delgado et al. (2010) was used todetermine the reddening E ( B − V ) and the logarithmof luminosity (log L ). We used a reddening law with R V = 3.1 to determine the absolute visual extinction A V .We adopted a distance d = 4000 ±
400 pc derived by Cu-sano et al. (2011) and Delgado et al. (2010) for cal-culating the luminosity. Two stars (J06445578+001368and J06450208+0019443) that do not have Delgado et al.(2010) optical photometry, archival B − V colors (Monetet al. 2003), and 2MASS J -band magnitudes were usedto derive the extinction and luminosity respectively. Weassumed that the effect of any infrared excess in the J -band is comparatively small. Table 2 summarizes thefundamental parameters determined for these stars us-ing SALT spectra.
VLT/VIMOS data
Additional data of PMS stars in Sh 2-284 was providedin Cusano et al. (2011). These authors obtained mul-tislit spectroscopy of ∼
900 objects covering a 30 × area (see Fig. 1) using the Visual & Multi-Object Spectrograph ( VIMOS ) on the Very Large Tele-scope (VLT). They identified 23 bonafide PMS stars fromthis sample based on H α emission, spectral type, andnear-infrared excess criteria. 17 show infrared-excess re-sembling Class II sources, whilst the remaining objectsresemble Class III sources (i.e. weakly accreting PMSstars). The location of these stars is shown in Fig. 1.Three stars (J06443682+0016186, J06450208+0019443,J06450681+0013535) also have SALT spectra.The spectral types and fundamental parameters for 22stars were determined by Cusano et al. (2011) usingthe method devised by Hern´andez et al. (2004). Onestar (J06452476+0013360) shows no absorption featureswhich means that it did not allow for spectral classi-fication. We adopt these results which are listed in Ta-ble 2. We compared the spectral parameters for the threestars in common with our SALT observations, and wefound that spectral types agreed within 2 sub-classes,which is the average error. The agreement is expectedas both spectral type determinations use similar line in-dicators and distances adopted by the authors are thesame as the values used in the analysis of the
SALT spectra. The logarithm of luminosity of one overlappingstar (J06443682+0016186) measured using near-infrared J -band magnitude is higher by 0.25 dex to the value mea-sured using Cusano et al. (2011) photometry. This isattributed to the near-infrared excess. For this star, weadopt the logarithm of luminosity determined by Cu-sano et al. For all other parameters determined usingboth SALT and
VIMOS spectra, we adopt the valuesestimated from the
SALT spectra.The overall sample consists of 24 bonafide PMS stars inSh 2-284 with spectral types ranging from mid A- earlyG for which there are observed parameters, which arenecessary to infer accretion properties. RESULTS
Masses and Ages
Using the measured effective temperatures and lumi-nosities, we place the PMS stars in an Hertzsprung- . . . . . . . . log T eff . . . . . . . l og L / L (cid:12) M (cid:12) M (cid:12) M (cid:12) M (cid:12) Fig. 4.—
Hertzsprung-Russell diagram of Sh 2-284 PMS stars.The solid vertical lines from right to left are Z = 0.004 PMSisochrones (Bressan et al. 2012) at 0.3, 1, 3 and 10 Myr, andthe zero-age main sequence respectively. The dashed horizontallines from bottom up are the 0.8, 1.5, 2, and 3 M (cid:12) tracks. Theaverage error bars in log T eff and log L are shown in the top-righthand corner. The log T eff and log L of the majority of our PMSstars have been taken from Cusano et al. (2011). See Table 3 forthe interpolated ages, and masses. Russell (H-R) diagram Fig. 4. Overlaid are the non-accreting single star isochrones and tracks calculated byBressan et al. (2012) at Z = 0.004. The masses andages are measured by interpolating their positions withrespect to tracks and isochrones, respectively. The de-rived masses and ages are listed in Table 3. The differ-ences that arise due to variations between different setsof tracks and isochrones are discussed in Appendix A.The mass range (0.9 - 2.6 M (cid:12) ) places a majority of thestars in the intermediate mass T Tauri or Herbig Ae typerange. These stars are very similar to accreting Classi-cal T Tauri stars and their ˙ M acc can be reliably mea-sured using their H α emission line luminosity (Calvet etal. 2004). The median error on the masses is . M (cid:12) .The ages determined from isochrones span a range of 1 -15 Myr. The PMS stars in the central cluster Dolidze 25have ages around 2 - 3 Myr, including the newly identifiedPMS stars using the SALT spectra (see Fig. 1). Thoselocated at the edges of the surrounding nebulous bubbles(see Puga et al. 2009 for a detailed description of thenomenclature) have younger ages, whereas those at therim have older ages. These results are in agreement withthe near-mid infrared study of Puga et al. (2009) andspectroscopic-photometric work of Cusano et al. (2011).Puga et al. (2009) interpret this as evidence that starformation was triggered by a previous generation of ion-izing stars in these regions, which has lead to the spreadof PMS ages inferred from the H-R diagram. Their ar-gument was supported by a first-principle analysis of thegas dynamics within the region.PMS membership of these stars is not suspect, as allobjects display infrared excesses resembling PMS stars,and there is little evidence for mid-A to early-G main se-quence stars having H α EW < − TABLE 2Physical parameters of PMS stars in Dolidze 25
ID (2MASS) Spectral EW (H α ) A V log T eff a log L b type (˚A) (mag) (K) ( L (cid:12) ) SALT spectraJ06443682+0016186 A9 − ± ± c J06445577+0013168 F2 − ± ± − ± ± − ± ± c J06450681+0013535 F9 − ± ± VIMOS spectra d J06443682+0016186 e A8V − ± e A5V − ± e G0V − ± − ± − ± − ± − ± − ± − ± − ± − ± − ± − ± − ± f G0V − ± − ± − ± − ± − ± − ± f K2V − ± − ± a Average error in log T eff a is around . SALT spectra, and 1subclass from
VIMOS spectra. b Average error in log L is ∼ d ∼
500 pc. c Log L was calculated using 2MASS J -band magnitude and archival (Monet et al. 2003) B − V colours. d Taken from Table 4 in Cusano et al. (2010), where the quoted average errors in spectraltype is one subclass, A V ∼ e Star also has
SALT longslit spectra. f Star not detected in 2MASS, coordinates from Cusano et al. (2010). been significantly more extincted than the other PMSstars. Moreover, one early G star (J06451318+0018307),with an estimated age >
10 Myr, also displays Li in ab-sorption in high signal-to-noise high resolution
VIMOS spectra. Li λ . ∼ M (cid:12) . This leads us to suggest that the agespread could possibly be due to errors in birthline cal-culations for G-type stars (Hartmann 2003), which maylead to a systematic overestimation of absolute stellarages. Therefore, any ages derived can only be accuratein a relative sense. Any such birthline effect is circum-vented by interpolating the ages of solar metallicity PMSstars used for comparison using the same isochrones andfollowing the same interpolation technique. This is de-scribed further in Appendix A. Accretion luminosity
In the current PMS evolutionary paradigm, stars ac-crete mass from a surrounding circumstellar disk. Thedisk is thought to be curtailed at some inner radius ( R in )by the stellar magnetosphere, and gas in the disk is ac-creted via the magnetic field lines. The gravitational en- V. M. Kalari & J. S. Vinkergy is released, as material falls onto the stellar surfacecreating an accretion shock, leading to excess continuumemission, which is particularly evident in the ultravioletwavelength range. The ionization and recombination ofgas during this process leads to strong Balmer line emis-sion (notably in H α ). In this scenario, the bolometricaccretion luminosity ( L acc ) can be estimated from there-radiated energy, which can be measured from the H α line luminosity ( L H α ).We measure L H α by adapting the method of Mohantyet al. (2005) (see also Costigan et al. (2014) for an ap-plication to higher-mass stars). The flux from Castelli &Kurucz (2004) model spectra (Z = 0.004) for the nearestpoint in ( T eff ,log g ) plane in 6554 - 6572 ˚A range is takenas continuum flux. We adopt for all stars log g = 4.0.The continuum flux is multiplied by the H α EW and4 πR to obtain L H α . Veiling is estimated by compar-ing with Pickles (1998) non-accreting stellar spectra. Inthe range with measurable absorption lines ( > L H α - L acc relation used by De Marchi et al. (2010);log L acc = log L H α + 1 . ± .
47 (1)The relation was derived by comparing L acc estimatedfrom the UV continuum-excess with L H α of 0.5 - 2 M (cid:12) mass stars (Dahm 2008). log L acc estimated this way in-cludes an uncertainty of factor 0.5, which is the dominanterror in the measured ˙ M acc . The L H α - L acc used by DeMarchi et al. (2010) agrees within errors with that in asimilar range derived by Mendigut´ıa et al. (2011). Westress that the L H α - L acc relation was adopted with a viewto reduce any sources of systematic errors between ourresults to those in the LMC. A comparison of the methodused to calculate L acc with respect to traditional H α lineflux measurements is presented in Appendix B. Mass accretion rate ˙ M acc can be estimated from the L acc using the free-fallequation; ˙ M acc = L acc R ∗ GM ∗ R in R in − R ∗ ! . (2)Here M ∗ , R ∗ are the stellar mass and radius respec-tively. The inner radius ( R in ) is uncertain and dependson the coupling of the accretion disk with the magneticfield lines. We follow Gullbring et al. (1998) and take R in = 5 ± R (cid:12) (Vink et al. 2005). Using the stellar massand radius measured from the H-R diagram, we estimatethe ˙ M acc for each star. The accretion properties are tab-ulated in Table 3.As a sanity check on the estimated ˙ M acc , we com-pared whenever possible with archival U -band photome-try of Delgado et al. (2010). ˙ M acc , or more directly the L acc can be measured reliably from U -band photometry.Comparison of the observed U -band magnitude with atemplate magnitude comprises a direct measure of theexcess U -luminosity caused by the emission shock whichcan be translated into L acc (Gullbring et al. 1998). − . − . − . − . − . H α log ˙ M acc ( M (cid:12) yr − ) − . − . − . − . − . U l og ˙ M a cc ( M (cid:12) y r − ) Fig. 5.—
Comparison of ˙ M acc measured from H α spectroscopy(abscissa) with archival U -band photometry (ordinate). The av-erage errors are shown in the top-left hand corner. Although the U -band sample is small, the comparison suggests consistency be-tween mass-accretion rates determined from H α and the U -band,i.e. H α is most likely a good indicator for mass-accretion phenom-ena. Five stars out of the total 24 have archive U -band pho-tometry in Delgado et al. (2010). The excess U -bandluminosity ( L U, excess ) of these stars, thought to be dueto accretion is estimated according to the formula L U, excess = 4 π W( d f obs − R ∗ f mod ) (3)(Romaniello et al. 2002). f mod and f obs are the modeland observed fluxes in the U -band respectively. W is thewidth of the U -band filter. The L U, excess is translatedinto L acc following the relation of Gullbring et al. (1998)log L acc = 1 .
09 log L U, excess + 0 . . (4)Using eq. (2), we compute ˙ M acc . A comparison of the˙ M acc derived using U -band photometry and H α line lu-minosity are presented in Fig. 5. We find that the dif-ference is not greater than the mean error (0.5 dex), sug-gesting that the ˙ M acc is reasonably well constrained forour purposes. Disk properties
We have shown that all 24 PMS stars in Sh 2-284 haveH α in emission, and for a subset the accretion rates de-rived from H α agree with those derived from the U-band(see also Kalari et al. 2014, in prep.). This demonstratesto a reasonable degree that these stars are accreting.Most of these stars also have near-infrared 2MASS JHKs magnitudes (Cutri et al. 2003), and
Spitzer imaging upto 8 µ m (Puga et al. 2009). In this section, we analyzethe disk properties of the PMS objects using infraredphotometry.The slope of the SED in the infrared, α IR = d log( λF λ ) d log λ , (5)at λ > µ m is used to diagnose the evolutionary stageof the disk-star system (Lada 1987). We adopt the clas-sification scheme of Greene et al. (1994) to distinguishre-main sequence accretion rates at low metallicity 7 TABLE 3Accretion and disc properties of PMS stars in Dolidze 25
No. ID (2MASS) Mass Age log L acc log ˙ M acc α IR Class( M (cid:12) ) (Myr) ( L (cid:12) yr − ) ( M (cid:12) yr − )1 J06443291+0023546 1.79 4.0 0.28 − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − between systems with protostellar disks (Class I), opti-cally thick disks (Class II), or no circumstellar excess(Class III). α IR was derived by fitting the Spitzer magni-tudes at 3.6, 4.5, 5.8, and 8 µ m. The resultant α IR valuesare given in Table 3. The disk SEDs are shown in Fig. 6.Five stars do not have photometry in any band. Forthese stars, we fitted the available Spitzer magnitudes tothe disk models of Robitalle et al. (2006) to classify thedisk evolutionary state.Out of the 19 accreting PMS stars that have
Spitzer photometry at all observed wavelengths, we find that 18have SED slopes resembling Class I/II objects, with onestar having α IR resembling a Class III object. Threestars do not have 8 µ m photometry (J06443788+0021509,J06444602+0019182), of which two have best-fit diskmodels suggestive of either flat or sources devoid of anyinfrared excess, and one (J06451701+0022077) sugges-tive of a dusty disk. Two stars that do not have pho-tometry at either 4.5, or 5.8 µ m (J06451701+0022238,J06450971+0014121), have best-fit models suggestive ofClass II objects. Overall, most of the PMS star samplehave infrared-excess akin to circumstellar disks, verifyingthat we deal with genuine accreting PMS stars. DISCUSSION
We present spectroscopic estimates of accretion ratesof 24 low- Z PMS stars in a stellar mass range between1 - 2 M (cid:12) , having a median age of 3.5 Myr, and a medianaccretion rate of 10 − . M (cid:12) yr − . The majority of oursample (21 out of 24 stars) shows evidence for circum-stellar disks, and all stars with literature U -band pho-tometry (5 out of 24) display an UV excess, confirmingthe PMS nature of the sample. Based on our results,we discuss in the following section the relation between L acc and stellar luminosity; and the observed dependenceof the estimated ˙ M acc on stellar mass and age. We alsocompare our results with literature estimates of accretion properties of Z (cid:12) Galactic and 1/3 Z (cid:12) LMC PMS stars toexamine if ˙ M acc estimates are a function of Z within theobserved Z range. L acc vs. L ∗ The detection limit on ˙ M acc are dependent on the de-tection limits of the measured L acc . This is demonstratedin the L acc vs. L ∗ distribution in Fig. 7. We find that L acc < L ∗ , i.e., we did not identify any continuum stars(Hern´andez et al. 2004) in our sample, leading to anupper boundary in the ˙ M acc - M ∗ plane, possibly leadingto a scarcity of high ˙ M acc . However, it is thought thatcontinuum stars form only a very small fraction of thetotal population in well-studied regions such as Taurus(Hartmann 2008).The lower detection limits have been calculated from L ∗ and M ∗ of stars lying on the 3 Myr isochrone with T eff = 5000, 6000, 7000, 8000, 10000 and 12000 K. Weassume a H α EW = − α line emission. We note that at L acc lower than this boundary it would be impossible todetect accretion using conventional means such as lineemission or UV-excess, since H α is the most sensitive in-dicator to tiny accretion rates. The lower boundary of L acc is within the error bars of detections at L ∗ < L (cid:12) ,at approximately 1.5 M (cid:12) . The PMS stars identified byCusano et al. (2011) which form the bulk of our sam-ple stars, were identified by the authors of that paperas those stars showing H α emission out of 1506 spec-tra observed in the region. Stars were rejected as non-members based on their luminosity classification, and in-frared colors. It is plausible to expect that the Cusano etal.(2011) PMS sample would contain stars having lower L acc at higher masses if they existed. This suggests thatour sample contains PMS stars at the lowest measurable V. M. Kalari & J. S. Vink . . . . . . log λ ( µ m) . . . . . . l og λ F λ ( e r g s c m − s − ) − . . . . − − − − . . . . − − − − . . . . − − − − . . . . − − − − . . . . − . − . − . − . . . . − . − . − . − . . . . − . − . − . − . . . . − . − . − . − . . . . − − − − . . . . − − − − . . . . − . − . − . − . . . . − − − − . . . . − − − − . . . . − − − − . . . . − . − . − . − . . . . − − − − . . . . − − − − . . . . − − − − . . . . − . − . − . − . . . . − − − − . . . . − − − − . . . . − . − . − . − . . . . − − − − . . . . − . − . − . Fig. 6.—
The spectral energy distribution of Sh 2-284 pre-main sequence stars. The number in the top-right corner corresponds to theindex number in Table 3. Dashed lines are the Kurucz & Castelli (2004) model spectra corresponding to the determined spectral parametersin Table 2. Dots are the un-reddened
U,B,V,R,I fluxes from Delgado et al. (2010) or Cusano et al. (2011). Crosses are
JHK s fluxes from2MASS (Cutri et al. 2003) and
Spitzer
IRAC fluxes at 3.6, 4.5, 5.8, and 8.0 µ m. re-main sequence accretion rates at low metallicity 9 . . . . . log L ( L ) . . . . . . . . l og L a cc ( L ) M M M M Fig. 7.— L acc as function of L ∗ . The dashed line displays the L ∗ = L acc relation. The inverted triangles give the detection limitscalculated for six different T eff for model L ∗ at 3 Myr. The upperleft corner displays the mean error bars. The L acc are given inTable 3. ˙ M acc within this range. One star lies at the boundary ofthe lower detection limit (J06450681+0013535) and hasa measured H α EW of − ˙ M acc as a function of stellar mass and age ˙ M acc are plotted as a function of M ∗ in Fig. 8. We findthe best-fit power law relationship across the 1 - 2 M (cid:12) range to be ˙ M acc ∝ M ∗ . ± . . There is a spread of ∼ M acc at any given M ∗ . The spread can partiallybe related to the intrinsic distribution of ages at anygiven mass, as ˙ M acc vary with age. But the scatter is alsofound in coeval populations, suggesting that there areother factors beyond mass and age controlling accretionrates (Hartmann et al. 2006).We employed survival analysis linear regression using ASURV (Lavalley et al. 1992) to include lower andhigher limits in accretion rates. We find that includ-ing lower limits leads to a steeper slope, but withinthe error bar. The inclusion of upper limits leads toa shallower slope, but within 0.1 dex of the slope calcu-lated. This suggests that the absence of low-mass high-accretion rates, or high-mass low-accretion rates is notstatistically significant in our regression fitting, and thatthe given distribution statistically represents a generaltrend of increasing ˙ M acc with M ∗ . The index α mea-sured is close to the Galactic median ∼
2, but deviantfrom the value of ∼ M acc -Age plane (Fig. 8). We fitthe orthogonal distance regression to account for errorson both ˙ M acc and t ∗ . We find the power-law index η to be − ± Z (cid:12) PMS stars, and as expected in line with viscousdisk evolution at these masses (Sicilia-Aguila et al. 2006;Manara et al. 2012).Overall, a comparison of Sh 2-284 PMS stars with theliterature accretion rates of 1/3 Z (cid:12) LMC and Z (cid:12) Galac- tic PMS stars is shown in Fig. 8. The Sh 2-284 PMSstars occupy similar ranges to Z (cid:12) PMS stars in boththe ˙ M acc - M ∗ and ˙ M acc - t ∗ plane. We find an absence ofLMC PMS stars at low ˙ M acc which suggest that the over-all slopes measured by Spezzi et al. (2012) may be sig-nificantly affected by detection limitations. This couldbe due to the difficulty of observing low-mass stars inthe distant LMC. We also find an absence of stars with˙ M acc > − M (cid:12) yr − at ages beyond 5 Myr in our sam-ple, a region of the plane that is populated by pointsfrom the LMC. Is ˙ M acc a function of Z ? We have shown that the ˙ M acc in Sh 2-284 and Galacticsolar- Z star forming regions follow similar distributionsin the ˙ M acc - M ∗ and ˙ M acc - t ∗ plane. The PMS star samplein the LMC of Spezzi et al. (2012) have shallower slopes,which may be affected by detection limits. However, asdiscussed by Spezzi et al. (2012), the median ˙ M acc at anygiven age in a defined mass range (1-2 M (cid:12) ) is a robusttool for comparing any differences in ˙ M acc between star-forming regions. To investigate this we tabulate the me-dian ˙ M acc of 1 - 2 M (cid:12) PMS stars in known star-formingregions in Table 4. The median age for the given sampleis computed from the isochronal ages. To estimate thecompleteness of each sample, we calculate the expectedfraction of stars within this mass range using the Kroupa(2001) IMF, and total cluster masses from Weidner &Kroupa (2006). The median age of each star-formingregion is used to determine the expected fraction of ac-cretion stars, N exp (Fedele et al. 2010).The median log ˙ M acc of Sh 2-284 PMS stars is − ∼ M acc reported bySpezzi et al. (2012) in an LMC PMS population having amedian age of 6 Myr is − M acc expected inthe SMC cluster NGC 602 at an age of 3 Myr for 1 - 2 M (cid:12) is expected to be & − M acc in Sh2-284 are comparable to Galactic values at similar ages.There is no systematic difference at an age of 3 Myr inthe accretion rates of solar Z or Sh 2-284 PMS starsbased on these results. If we assume that our results arenot significantly underestimated (see Section 4.2), it ischallenging to explain physically how accretion is pro-ceeding at much higher rates in much older stars in theLMC, especially considering that Z LMC ∼ × Z Sh 2 − .It seems unlikely that Z is the only quantity determin-ing the accretion rates, and factors other than Z may beresponsible for the high ˙ M acc measured by the studies ofDe Marchi et al. (2010; 2011a;2011b;2013) and Spezzi etal. (2012).There are practical considerations which might lead tohigher measured ˙ M acc at low Z . The observable limitfor the highest measurable ˙ M acc is generally given by L acc < L ∗ , owing to the difficulty in detecting continuum stars, and the lowest ˙ M acc at H α EW < A beyond which0 V. M. Kalari & J. S. Vink − .
05 0 .
00 0 .
05 0 .
10 0 .
15 0 .
20 0 .
25 0 . log M ∗ ( M (cid:12) ) − . − . − . − . − . − . − . l og ˙ M a cc ( M (cid:12) y r − ) . . . . . log t (yr) − . − . − . − . − . − . − . − . l og ˙ M a cc ( M (cid:12) y r − ) M (cid:12) Distribution of ˙ M acc as function of M ∗ (left) and t ∗ (right) shown as blue circles. The solid line is the best-fit regression slopein the 1 - 2 M (cid:12) range, and has an index α = 2.4 ± M acc - M ∗ plane and index η = 0.7 ± M acc - t ∗ plane. The dashed line(right) is the expected evolution of viscous disks following Hartmann et al. (1998), and left is the slope for LMC PMS stars reported inSpezzi et al. (2012). Black inverted triangles are the detection limits. Triangles are the data points for Galactic PMS taken from thestudies of Natta et al. (2006); Herczeg & Hillenbrand (2008); Sicilia-Aguilar et al. (2010); Barentsen et al. (2011; 2013); Antoniucci et al.(2011); Manara et al. (2012); Alc´ale et al. (2014). Red crosses are taken from the studies of LMC PMS stars from Spezzi et al. (2012). TABLE 4Median ˙ M acc of PMS stars in the 1 -2 M (cid:12) at different Z Region N N exp Age log ˙ M acc Reference(Myr) ( M (cid:12) yr − ) Z = Z (cid:12) L1641 23 - 1 − ± 10 1.5 − ρ -Ophiuchus 5 5 ± a − ± − ± − ± b − − − − Z = 0.4 Z (cid:12) LMC Field3 Pop.2 146 6 − − − − − Z = 0.2 Z (cid:12) Sh 2-284 20 3.5 − Z = 0.1 Z (cid:12) NGC 346 d - 1 − d - 20 − Note . — (a) ρ -Ophiuchus is oft-quoted with an age of 0.5 Myr (Natta et al. 2006), however theages of the accreting PMS stars are thought to be as old as 2 Myr (Wiliking et al. 2008; Rigiliaco etal. 2010). (b) The total cluster mass of Trumpler 37 is debated, as studies of stellar content reveal adearth of near solar mass members, and a deviant stellar mass function slope (Errman et al. 2012).(c) The age of IC 348 is debated, with initial studies suggesting an age as young as 1.5 Myr (Herbig1998), but later revised to around 4 Myr (Mayne & Naylor 2007). (d) Taken from Figure 8 of Spezziet al. (2012). re-main sequence accretion rates at low metallicity 11accretion is hard to measure. The lower limit leads to athreshold in L ∗ (shown in Fig. 7). L ∗ increases inverselyin a non-monotonic fashion as a function of Z accordingto stellar isochrones. This suggests that we are measur-ing stars of similar spectral type, but with different L ∗ atdifferent Z , with consequently higher L acc and thereby˙ M acc at lower Z . This also leads to raising the L acc - L ∗ upper limit shown in Fig. 7, and leading to a slightlyhigher ˙ M acc . To test the difference that may be causedby this effect, we calculate the lower detection limit us-ing solar Z isochrones and use these to measure themedian ˙ M acc for censored data using a Kapalan-Meiertest. We find that this leads to a slightly lower median˙ M acc ∼ − Z .This work is based on the assumption that the metallic-ity of the PMS stars in Sh2-284 is similar to that of the Bstars located in the central open cluster Dolidze 25. Themetallicity of Dolidze 25 is anomalous when taken intoaccount the distance-metallicity gradient of the Galaxy(Rolleston et al. 2000). It has been suggested that theclose association with the Canis Major Dwarf Galaxy isthe reason for the metal-deficient nature of the centralOB stars (Martin et al. 2004). It seems reasonable toassume that the metallicity of the PMS stars in the Sh2-284 region is similar to the central B stars, however, ifother processes lead to the metal-deficient nature of thecentral B stars, the PMS stars may have higher abun-dances. Future work on the Z determination of PMSstars in Sh2-284 would thus be welcomed. SUMMARY AND FUTURE WORK We presented an optical spectroscopic study of PMSstars located in the star-forming region Sh 2-284. Sh2-284 is a Galactic star-forming region with Z ∼ Z (cid:12) . Our sample consists of 24 objects span-ning a mass range between 0.9 - 2.6 M (cid:12) and an age rangebetween 1 - 10 Myr, with a median age of 3.5 Myr. Ourresults are consistent with the scenario of sequential starformation suggested by Puga et al. (2009) inferred fromgas dynamics. 21 stars have significant infrared excessup to 8 µ m, suggesting that they are bonafide Class I/IIPMS stars, whilst the remaining PMS stars are mostlikely Class III objects.We used the measured H α EW, and stellar parametersto derive L H α , from which we estimated the ˙ M acc . Forfive objects, we also have archival U -band photometry,and we measured the U -band excess intensity from whichwe estimated ˙ M acc . We found that the ˙ M acc estimatedfrom the L H α agree well with those derived from the U -band excess. The multiple signatures of disk accretionare highly suggestive of ongoing accretion in the Sh 2-284 PMS stars.Our sample is detection limited in L acc and ˙ M acc . Thehighest measured L acc is never greater than L ∗ , and thelowest measured L acc corresponds to a H α EW ∼ − M acc to steeply increase with M ∗ , with a power-law index2.4 ± M (cid:12) range. While ˙ M acc decrease with stellar age, with anindex − ± ∼ Z PMS stars, we findno significant differences in the 1 - 2 M (cid:12) range. The dis-tribution of sources in the ˙ M acc - M ∗ and ˙ M acc - t ∗ planesare very similar, although there are fewer lower ˙ M acc ob-jects at younger ages in Sh 2-284. However, this may bea product of low number statistics at very young ages.The median log ˙ M acc of Sh 2-284 ( − M acc of 3 Myr old solar- Z starforming regions. When compared to the results in theLMC ( Z = 0.008) from Spezzi et al. (2012), we find thatthe median log ˙ M acc of LMC PMS stars are ∼ − M acc are much lowerin Sh 2-284. We also find that the distribution of Sh 2-284sources in in the ˙ M acc - M ∗ and ˙ M acc - t ∗ planes occupy re-gions at lower ˙ M acc when compared to LMC stars. Thisdifference is not easily explained, particularly when weconsider that the Z of the LMC is twice that of Sh 2-284,and if Z is truly proportional to ˙ M acc , ˙ M acc of Sh 2-284PMS stars would be similarly higher than the LMC aver-age, as the LMC measures are compared to the Galaxy.It is likely that the Spezzi et al. (2012) results are influ-enced by completeness issues. Moreover, it is not clearwhether accretion can be sustained at such high rates forsuch long periods as measured in the LMC PMS stars,which, if true would suggest different disk evolutionarymechanisms in comparison to Galactic PMS stars, whereaccretion is thought to cease by around 5 Myr. We sug-gest that there are most likely factors beyond differencesin Z which influence the measured differences in ˙ M acc .These include practical differences in detection limits atdifferent Z , and difficulty in identifying PMS stars atdistances to the LMC. A more systematic and thoroughstudy of PMS stars in the Magellanic Clouds and theGalaxy is required to disentangle the various factors thatinfluence measured mass-accretion rates.Based on observations obtained with the South AfricanLarge Telescope under ID 2013-2-UKSC-008; and theESO telescopes at the La Silla Paranal Observatory un-der program no. 0.74.C-0111. We would like to thank thereferee and Drs. Antonella Natta and Danny Lennon foruseful comments. V.M.K. is the recipient of a postgradu-ate scholarship awarded by the Department for Culture,Arts and Leisure Northern Ireland. J.S.V. and V.M.K.acknowledge funding from the the Science and Technol-ogy Facilities Council U.K. REFERENCESAlcal´a, J. M., Natta, A., Manara, C. F., et al. 2014, A&A, 561,AA2Antoniucci, S., Garc´ıa L´opez, R., Nisini, B., et al. 2011, A&A,534, AA32 Barentsen, G., Vink, J. S., Drew, J. E., & Sale, S. E. 2013,MNRAS, 429, 1981Barentsen, G., Vink, J. S., Drew, J. E., et al. 2011, MNRAS, 415,103 Beaulieu, J. P., Lamers, H. J. G. L. M., Grison, P., et al. 1996,Science, 272, 995Bressan, A., Marigo, P., Girardi, L., et al. 2012, MNRAS, 427, 127Burgh, E. B., Nordsieck, K. H., Kobulnicky, H. A., et al. 2003,Proc. SPIE, 4841, 1463Calvet, N., Muzerolle, J., Brice˜no, C., et al. 2004, AJ, 128, 1294Cai, K., Durisen, R. H., Michael, S., et al. 2006, ApJ, 636, L149Caratti o Garatti, A., Garcia Lopez, R., Antoniucci, S., et al.2012, A&A, 538, AA64Castelli, F., & Kurucz, R. L. 2004, arXiv:astro-ph/0405087Costigan, G., Vink, J. S., Scholz, A., Ray, T., & Testi, L. 2014,MNRAS, 440, 3444Cusano, F., Ripepi, V., Alcal´a, J. M., et al. 2011, MNRAS, 410,227Cutri, R. M., Skrutskie, M. F., van Dyk, S., et al. 2003, VizieROnline Data Catalog, 2246, 0Dahm, S. E. 2008, AJ, 136, 521De Marchi, G., Beccari, G., & Panagia, N. 2013, ApJ, 775, 68De Marchi, G., Paresce, F., Panagia, N., et al. 2011a, ApJ, 739, 27De Marchi, G., Panagia, N., Romaniello, M., et al. 2011b, ApJ,740, 11De Marchi, G., Panagia, N., & Romaniello, M. 2010, ApJ, 715, 1de Wit, W. J., Beaulieu, J. P., Lamers, H. J. G. L. M., Coutures,C., & Meeus, G. 2005, A&A, 432, 619de Wit, W. J., Beaulieu, J.-P., Lamers, H. J. G. L. M., Lesquoy,E., & Marquette, J.-B. 2003, A&A, 410, 199de Wit, W. J., Beaulieu, J. P., & Lamers, H. J. G. L. M. 2002,A&A, 395, 829Delgado, A. J., Djupvik, A. A., & Alfaro, E. J. 2010, A&A, 509,A104Duncan, D. K., & Jones, B. F. 1983, ApJ, 271, 663Ercolano, B., & Clarke, C. J. 2010, MNRAS, 402, 2735Fang, M., van Boekel, R., Wang, W., et al. 2009, A&A, 504, 461Favata, F., Micela, G., Sciortino, S., & D’Antona, F. 1998, A&A,335, 218Fedele, D., van den Ancker, M. E., Henning, T., Jayawardhana,R., & Oliveira, J. M. 2010, A&A, 510, A72Fitzsimmons, A., Dufton, P. L., & Rolleston, W. R. J. 1992,MNRAS, 259, 489Greene, T. P., Wilking, B. A., Andre, P., Young, E. T., & Lada,C. J. 1994, ApJ, 434, 614Gullbring, E., Hartmann, L., Brice˜no, C., & Calvet, N. 1998,ApJ, 492, 323Hartmann, L. 2008, Accretion Processes in Star Formation, byLee Hartmann, Cambridge, UK: Cambridge University Press,2008,Hartmann, L., D’Alessio, P., Calvet, N., & Muzerolle, J. 2006,ApJ, 648, 484Hartmann, L. 2003, ApJ, 585, 398Hartmann, L. 2001, AJ, 121, 1030Hartmann, L., Calvet, N., Gullbring, E., & D’Alessio, P. 1998,ApJ, 495, 385Hartmann, L., Cassen, P., & Kenyon, S. J. 1997, ApJ, 475, 770Herczeg, G. J., & Hillenbrand, L. A. 2008, ApJ, 681, 594Hern´andez, J., Calvet, N., Brice˜no, C., Hartmann, L., & Berlind,P. 2004, AJ, 127, 1682 Kalari, V. M., Vink, J. S., Dufton, P. L., et al. 2014, A&A, 564,L7Kenyon, S. J., & Hartmann, L. 1995, ApJS, 101, 117Kroupa, P. 2001, MNRAS, 322, 231Lada, C. J. 1987, Star Forming Regions, 115, 1Lamers, H. J. G. L. M., Beaulieu, J. P., & de Wit, W. J. 1999,A&A, 341, 827Lavalley, M., Isobe, T., & Feigelson, E. 1992, Astronomical DataAnalysis Software and Systems I, 25, 245Lennon, D. J., Dufton, P. L., Fitzsimmons, A., Gehren, T., &Nissen, P. E. 1990, A&A, 240, 349Manara, C. F., Robberto, M., Da Rio, N., et al. 2012, ApJ, 755,154Martin, N. F., Ibata, R. A., Bellazzini, M., et al. 2004, MNRAS,348, 12Mendigut´ıa, I., Calvet, N., Montesinos, B., et al. 2011, A&A, 535,A99Mohanty, S., Basri, G., & Jayawardhana, R. 2005, AstronomischeNachrichten, 326, 891Monet, D. G., Levine, S. E., Canzian, B., et al. 2003, AJ, 125, 984Muzerolle, J., Luhman, K. L., Brice˜no, C., Hartmann, L., &Calvet, N. 2005, ApJ, 625, 906Natta, A., Testi, L., & Randich, S. 2006, A&A, 452, 245Parker, Q. A., Phillipps, S., Pierce, M. J., et al. 2005, MNRAS,362, 689Palla, F., & Stahler, S. W. 1999, ApJ, 525, 772Palla, F., & Stahler, S. W. 1990, ApJ, 360, L47Panagia, N., Romaniello, M., Scuderi, S., & Kirshner, R. P. 2000,ApJ, 539, 197Pickles, A. J. 1998, PASP, 110, 863Pogodin, M. A., Hubrig, S., Yudin, R. V., et al. 2012,Astronomische Nachrichten, 333, 594Pringle, J. E. 1981, ARA&A, 19, 137Puga, E., Hony, S., Neiner, C., et al. 2009, A&A, 503, 107Robitaille, T. P., Whitney, B. A., Indebetouw, R., Wood, K., &Denzmore, P. 2006, ApJS, 167, 256Rolleston, W. R. J., Smartt, S. J., Dufton, P. L., & Ryans,R. S. I. 2000, A&A, 363, 537Romaniello, M., Robberto, M., & Panagia, N. 2004, ApJ, 608, 220Romaniello, M., Panagia, N., Scuderi, S., & Kirshner, R. P. 2002,AJ, 123, 915Sharpless, S. 1959, ApJS, 4, 257Sicilia-Aguilar, A., Henning, T., & Hartmann, L. W. 2010, ApJ,710, 597Sicilia-Aguilar, A., Hartmann, L. W., F¨ur´esz, G., et al. 2006, AJ,132, 2135Spezzi, L., de Marchi, G., Panagia, N., Sicilia-Aguilar, A., &Ercolano, B. 2012, MNRAS, 421, 78Stahler, S. W. 1983, ApJ, 274, 822Tognelli, E., Prada Moroni, P. G., & Degl’Innocenti, S. 2011,A&A, 533, A109Tout, C. A., Livio, M., & Bonnell, I. A. 1999, MNRAS, 310, 360Vink, J. S., Drew, J. E., Harries, T. J., Oudmaijer, R. D., &Unruh, Y. 2005, MNRAS, 359, 1049Weidner, C., & Kroupa, P. 2006, MNRAS, 365, 1333Yasui, C., Kobayashi, N., Tokunaga, A. T., Saito, M., & Tokoku,C. 2009, ApJ, 705, 54APPENDIX AGES AND ˙ M ACC DERIVED USING DIFFERENTSTELLAR TRACKS AND ISOCHRONES We use the PMS isochrones and tracks from Bressanet al. (2012) to derive the ages and masses for our sam-ple. To account for differences in stellar models whencomparing our results with literature results, we esti-mate the ˙ M acc , M ∗ , and t ∗ using the same Bressan etal. (2012) stellar tracks and isochrones at the appropri-ate Z . We used the quoted log L , and T eff of PMS starsin the LMC studies of Spezzi et al. (2012), and the OrionNebula study by Manara et al. (2012) to estimate M ∗ and t ∗ using the Bressan et al. isochrones and tracks. The L acc quoted by the Spezzi et al. (2012) and Manaraet al. (2012) study were combined with the appropriate M ∗ to derive ˙ M acc using Equation 2. In Figs. 9 and 10,we compare the ˙ M acc and ages derived using the Bres-san et al. (2012) stellar models with those quoted in thestudies. Namely, Spezzi et al. (2012) employed the Pisaisochrones database (Tognelli et al. 2011); while Manaraet al. (2012) derived results using a variety of isochrones,where we chose the results they derived from employingthe Palla & Stahler isochrones (1999). We find small sys-tematic offsets in the ages and ˙ M acc when compared tothe results of Spezzi et al. (2012), but no systematic dif-ferences in ˙ M acc compared to Manara et al. (2012). Wefind a non-systematic change in the ages of Orion Nebulare-main sequence accretion rates at low metallicity 13 . . . . . . . . logAge (yr) this work . . . . . . . . l og A g e ( y r) f r o m Sp ezz i e t a l. − . − . − . − . − . − . − . − . − . − . log ˙ M acc ( M (cid:12) yr − ) this work − . − . − . − . − . − . − . − . − . − . l og ˙ M a cc ( M (cid:12) y r − ) f r o m Sp ezz i e t a l. Fig. 9.— Comparison of accretion properties estimated using stellar parameters of PMS stars in the LMC (from Spezzi et al. 2012).Left: A comparison of the ages estimated by Spezzi et al. (2012) using Tognelli et al. (2011) isochrones (ordinate) and in this work usingBressan et al. (2012) isochrones (abscissa) at the appropriate Z . Right: A comparison of ˙ M acc derived using masses, radii using Bressanet al. (2012) isochrones (abscissa) versus those derived by Spezzi et al. (2012) (ordinate). . . . . . . . . logAge (yr) this work . . . . . . . . l og A g e ( y r) f r o m M a n a r a e t a l. − − − − − − log ˙ M acc ( M (cid:12) yr − ) this work − − − − − − l og ˙ M a cc ( M (cid:12) y r − ) f r o m Sp ezz i e t a l. Fig. 10.— Comparison of accretion properties estimated using stellar parameters of PMS stars in the Orion Nebula (from Manara et al.2012). Left: A comparison of the ages estimated by Manara et al. (2012) using Palla & Stahler (1999) isochrones (ordinate) and in thiswork using Bressan et al. (2012) isochrones (abscissa). Right: A comparison of ˙ M acc derived using masses, radii from Bressan et al. (2012)isochrones, and L H α - L acc relation used in this work (abscissa) versus those derived by Manara et al. (2012) (ordinate). The L acc used toderive the ˙ M acc was taken from Manara et al. (2012). PMS stars less than 1 Myr using different isochrones. TESTING THE ˙ M ACC ESTIMATION METHOD We demonstrate that the method used to calculate˙ M acc reproduce the ˙ M acc estimated from a direct mea-surement of the H α line flux in Fig. 11. We used theH α EW and stellar parameters ( T eff , M ∗ , R ∗ , A V , d )of eight Herbig Ae/Be stars determined by Pogodin etal. (2012), to calculate their ˙ M acc using the method de-scribed in Section 3.2 and 3.3. We compare the ˙ M acc cal-culated this way, with those estimated from the intensityof H α emission line measured from flux-calibrated spec-tra by Pogodin et al. (2012). We find that our resultsaccurately reproduce the measured ˙ M acc within 0.15 dex,the error due to uncertainties in stellar parameters andmeasured H α EW. − . − . − . − . M acc this paper ( M (cid:12) yr − ) − . − . − . − . l og ˙ M a cc P ogo d i n e t a l. ( M (cid:12) y r − ) Fig. 11.— Comparison of ˙ M accacc