The Effelsberg survey of FU~Orionis and EX~Lupi objects II. -- H$_2$O maser observations

FU Orionis (FUor) and EX Lupi (EXor) type objects are two groups of peculiar and rare pre-main sequence low-mass stars that are undergoing powerful accretion outbursts during their early stellar evolution. Water masers are widespread in star forming regions and are powerful probes of mass accretion and ejection, but little is known about the prevalence of them toward FUors/EXors. We perform the first systematic search for the 22.2 GHz water maser line in FUors/EXors to determine its overall incidence to perform follow-up high angular resolution observations. We used the Effelsberg 100-m radio telescope to observe the 22.2 GHz H2O maser toward a sample of 51 objects. We detect 5 water masers; 3 are associated with eruptive stars, resulting in a 6% detection rate for eruptive sources. These detections include one EXor, V512 Per (also known as SVS 13 or SVS 13A), and two FUors, Z CMa and HH 354 IRS. This is the first reported detection of water maser emission towards HH 354 IRS. We detect water maser emission in our pointing towards the FUor binary RNO 1B/1C, which most likely originates from the nearby deeply embedded source IRAS 00338+6312 (~4'', from RNO 1B/1C). Emission was also detected from H$_2$O(B) (also known as SVS 13C), a Class 0 source ~30'', from the EXor V512 Per. The peak flux density of H$_2$O(B) in our observations, 498.7 Jy, is the highest observed to date. In addition to the two non-eruptive Class 0 sources (IRAS 00338+6312 and H$_2$O(B) /SVS 13C), we detect maser emission towards one Class 0/I (HH 354 IRS) and two Class I (V512 Per and Z CMa) eruptive stars. We demonstrate the presence of 22.2 GHz water maser emission in FUor/EXor systems, opening the way to radio interferometric observations to study these eruptive stars on small scales. Comparing our data with historical observations suggest that multiple water maser flares have occurred in both V512 Per and H$_2$O(B).


Introduction
Low-mass young stellar objects (YSOs) are stars in the early stages of stellar evolution, specifically protostars and pre-main sequence (PMS) stars, which can undergo accretion-driven episodic outbursts. Studies of outbursting objects provide crucial information on the formation and the evolution of Sun-like stars. Amongst PMS stars, there are two small, but rather spectacular classes of outbursting low-mass YSOs: FU Orionis and Member of the International Max Planck Research School (IM-PRS) for Astronomy and Astrophysics at the Universities of Bonn and Cologne.
EX Lupi-type stars (FUors and EXors for short, respectively). Members of both classes show major increases in their optical and near-infrared (NIR) brightnesses. FUors can brighten by up to 5 -6 magnitudes in the optical, triggered by enhanced accretion from the accretion disk onto the protostar (Hartmann & Kenyon 1996;Herbig 1989). This phase can last for several decades, or even centuries (e.g. the recent review by Fischer et al. 2022, and references therein). For example, the prototype of the FUor class, FU Orionis, went into outburst in 1936 (Wachmann 1954), and remains in a highly active state. After a few other objects were observed to experience similar outbursts, Herbig (1977) defined the FUor class, which continues to increase in Article number, page 1 of 12 arXiv:2305.00736v1 [astro-ph.SR] 1 May 2023 size as new FUor-type objects are identified (e.g., Audard et al. 2014;Szegedi-Elek et al. 2020) and currently contains more than a dozen objects. The EXor class was defined by Herbig (1989), based on the properties of the prototype star EX Lupi, and currently also includes more than a dozen objects (e.g., Audard et al. 2014;Park et al. 2022). EXors can brighten by up to 1 -5 magnitudes in the optical and remain in a bright state for a few months or a few years (see e.g., Jurdana-Šepić et al. 2018); furthermore, their outbursts are recurring (e.g., Audard et al. 2014;Cruz-Sáenz de Miera et al. 2022).
Interstellar masers are powerful tools for studying the physics of star formation on small scales, frequently probing regions of enhanced density and temperature (e.g., Elitzur 1992;Reid & Honma 2014). While masers have been substantially used to probe both low-and high-mass star formation regions (e.g., Abraham et al. 1981;Omodaka et al. 1999;Hirota et al. 2011;Furuya et al. 2001Furuya et al. , 2003, so far little information exists on masers in FUors/EXors. Pioneering studies found compact maser emission in the 1720 MHz hyperfine structure line of hydroxyl (OH) toward the archetypal FUor V1057 Cyg (Lo & Bechis 1973). This emission, which comes from the immediate vicinity of the star (Lo & Bechis 1974) and is highly time variable (Winnberg et al. 1981), is unique in the literature. The 22.2 GHz transition of water (H 2 O) is the most widespread interstellar maser (see, e.g., Gray 2012, and references therein). It has been detected towards numerous low-to high-mass star forming regions in the Milky Way (see e.g. Ladeyschikov et al. 2022). Pumping models indicate that 22.2 GHz water masers are excited at elevated temperatures (∼500 K) and densities (10 8−9 cm −3 ), which are typically found in the compressed postshock regions of jets/outflows from YSOs (Elitzur et al. 1989a;Elitzur & Fuqua 1989;Gray 2012;Gray et al. 2022). With verylong-baseline interferometry (VLBI), multi-epoch observations of water masers associated with protostellar outflows can be used to study mass accretion and ejection (see, for example, Burns et al. 2016;Moscadelli et al. 2019). This suggests that water masers could potentially serve as valuable probes of mass accretion and ejection in FUors/EXors.
Despite the fact that water masers are closely associated with mass accretion and ejection in protostars, a systematic search for 22.2 GHz H 2 O masers in FUors/EXors has not yet been performed. Hence, the overall incidence of 22.2 GHz water masers in these classes of eruptive objects is unknown. In this paper, we present the first dedicated 22.2 GHz water maser survey of low-mass young eruptive stars, using the Effelsberg 100-m telescope. Our single-dish survey is a first step in investigating water masers in low-mass outbursting systems, aimed at investigating the existence and prevalence of water masers in these objects and identifying targets for follow-up interferometric observations. This paper is the second in a series (the first being Szabó et al. 2023) presenting radio and (sub)millimeter observations of FUors and EXors and their natal environments, and is organized as follows. In Sect. 2, we summarize our observations. In Sect. 3, we present our results, focusing on sources with water maser detections. In Sect. 4, we discuss our results, and in Sect. 5 we summarize our most important findings.

Observations
The H 2 O J K a ,K c = 6 16 − 5 23 transition (rest frequency 22235.0798 MHz, from the JPL Molecular Spectroscopy database 1 , Pickett et al. 1998) was observed simultaneously 1 https://spec.jpl.nasa.gov/ with the three lowest metastable NH 3 transitions ((J, K) = (1, 1), (2, 2) and (3, 3)), which were presented in Paper I (Szabó et al. 2023). The observations were carried out on 2021 November 18, November 23, and 2022 January 25 using the Effelsberg 100-m telescope in Germany 2 (project id: 95-21, PI: Szabó). The sample consisted of 51 sources: 33 FUors, 13 EXors, and 5 Gaia alerts. Gaia alert sources were chosen from the variable sources identified by the Gaia Photometric Science Alerts system (Hodgkin et al. 2021) based on light curve characteristics and luminosities similar to those of FUors/EXors. Five Gaia alert sources in our sample are yet to be classified; one source, Gaia18dvy, is listed with its Gaia alert name (Table B.1) but counted as a FUor based on its classification by Szegedi-Elek et al. (2020).
Our observations were performed in position-switching mode with an off-position at an offset of 5 east of our targets in azimuth. During our observations, the 1.3 cm double beam and dual polarization secondary focus receiver was employed as the frontend, while the Fast Fourier Transform Spectrometers (FFTSs) were used as the backend. Each FFTS provides a bandwidth of 300 MHz and 65536 channels, which gives a channel width of 4.6 kHz, corresponding to a velocity spacing of 0.06 km s −1 at 22.2 GHz. The actual spectral resolution is coarser by a factor of 1.16 (Klein et al. 2012).
At the beginning of each observing session, pointing and focus were verified towards NGC 7027. On 2021 November 18 we also targeted W75N, known for its H 2 O and NH 3 emission, to make sure that the system was working properly (see Appendix A). Pointing was regularly checked on nearby continuum sources, and was found to be accurate to about 5 . NGC 7027 was also used as our flux calibrator, assuming a flux density of ∼5.6 Jy at 22.2 GHz (Ott et al. 1994). The on-source integration time was 2.5 minutes per spectrum, and during each observing epoch, 4 spectra per source were obtained.
The majority of our sources were observed on 2021 November 18 and 23 (see Tables 2 and B.1). On 2021 November 18, we detected H 2 O maser emission toward V512 Per (SVS 13A), RNO 1B/1C, and HH 354 IRS. To study the time variability of the maser emission, we re-observed detected sources in as many subsequent epochs as possible (see Table 2), within the constraints of our allocated observing sessions. For Z CMa, which was known to have water maser emission (Moscadelli et al. 2006) but could not be observed in November 2021 due to time constraints, we searched for short-term maser variability by observing this source for two 4×2.5 minute blocks separated by 2.5 hours in January 2022. No variability was detected on this timescale, so all 8 spectra of Z CMa were averaged for the subsequent analysis. We note, that due to the weak detection of the water maser in HH 354 IRS, the spectrum was spectrally smoothed by a factor of 2 using the smooth built-in function in CLASS. The smoothed spectrum is presented throughout this paper. Having detected unusually high-amplitude (factor of ∼4 with respect to the previous observation) and rapid variability in the H 2 O maser spectra towards V512 Per (SVS 13A) (see Sect. 3.2.1), we also carried out nine-point observations and 1 ×1 On-The-Fly (OTF) mapping of this source on 2022 February 5 to investigate whether emission from nearby sources in the telescope sidelobes could be contributing to the observed emission. Consequently, we serendipitously detected strong water maser emission toward H 2 O(B) (SVS 13C), which is 30 from V512 Per (SVS 13A) (see Sect. 3.2.1 and 3.3.1). We also performed single-pointing observations towards H 2 O(B) during this epoch. We adopted the method introduced by Winkel et al. (2012) for our spectral calibration which resulted in a calibration uncertainty of about 15%. The half-power beam width (HPBW) was about 40 at 22 GHz and the main beam efficiency was 60.2% at 22 GHz. The conversion factor from flux density, S ν , to main beam brightness temperature, T mb , was T mb /S ν =1.73 K/Jy. Typical RMS noise levels for observations of detected sources are given in Table 2 and 3σ upper limits for non-detections are given in Table B.1.
The data were reduced using the GILDAS/CLASS package developed by the Institut de Radioastronomie Millimétrique (IRAM) 3 (Pety 2005;Gildas Team 2013). For each target, spectra observed on the same day were averaged to improve the signal-to-noise ratio prior to subtracting a linear baseline. Velocities are presented with respect to the local standard of rest (LSR) throughout this paper.

Results
Of our 51 targets, we detected >3σ water maser emission towards two FUors (Z CMa and HH 354 IRS) and one EXor (V512 Per/SVS 13A), corresponding to a detection rate of ∼6% towards eruptive stars. We also serendipitously detected water maser emission towards two non-eruptive embedded protostars, which we discuss in Sects. 3.3.1 and 3.3.2. The basic parameters of sources with maser detections, including types, coordinates, distances, and evolutionary classifications are listed in Table 1. In all, we detected water masers in two non-eruptive Class 0 sources (IRAS 00338+6312 and H2O(B)/SVS 13C) and in one Class 0/I (HH 354 IRS) and two Class I (V512 Per/SVS 13A and Z CMa) eruptive objects, using the standard classification scheme (see, e.g., Greene et al. 1994;Evans et al. 2009).
For sources with water maser detections, we fitted each velocity component with a Gaussian to obtain its LSR velocity ( LSR ), line width (∆ ), and peak flux density (S ν ), given in Table 2. The peak flux densities of detected water masers vary from 0.11 Jy to 498.7 Jy, spanning over 3 orders of magnitude. The observed maser velocities are within 10 km s −1 of the systemic cloud velocities measured from NH 3 emission. While shock velocities of 50 km s −1 are expected in theoretical models (e.g., Elitzur et al. 1989b), the modest velocity offsets between water masers and dense gas observed in our sample are generally consistent with observations of water masers towards high-mass YSOs (e.g., Urquhart et al. 2009;Cyganowski et al. 2013, Fig. 4 and Fig. 16 respectively). Isotropic H 2 O maser luminosities, L H 2 O , were calculated as (e.g., Anglada et al. 1996;Urquhart et al. 2011;Cyganowski et al. 2013): where D is the distance to the target (see Table 1). Estimating the isotropic H 2 O maser luminosities of individual velocity components separately, we find a range of L H 2 O of 7.9×10 −10 L to 6.1×10 −7 L (see Table 2). In the following subsections, we discuss our results for sources with detected water masers. Our non-detections are presented in Appendix B, where Table B.1 lists the targeted sources along with their types, coordinates, 3σ upper limits, whether they were previously searched for 22.2 GHz maser emission and 3 https://www.iram.fr/IRAMFR/GILDAS/ if so the reference, the date of observation in the current survey, their classification and reference, and distances.
For 31 sources in our sample, no previous observations of the 22.2 GHz water maser line have been reported in the literature.

Z CMa
Z CMa consists of an FUor (southwest component) and a Herbig Ae/Be star (northeast component) that are only 0.1 apart (Koresko et al. 1991;Bonnefoy et al. 2017). Figure 1 shows the H 2 O maser spectrum observed toward Z CMa, the only source among those detected observed at only one epoch (Sect. 2). As shown in Figure 1, there is only one bright maser feature, at LSR =7.82 km s −1 , blueshifted by ∼6 km s −1 with respect to the thermal NH 3 emission. Although Z CMa has been observed in many previous water maser studies (Blitz & Lada 1979;Thum et al. 1981;Deguchi et al. 1989;Scappini et al. 1991;Palla & Prusti 1993;Moscadelli et al. 2006;Sunada et al. 2007;Bae et al. 2011;Kim et al. 2018
As shown in Figure 2, we detected weak H 2 O maser emission (peak flux densities <0.2 Jy, Table 2) towards HH 354 IRS in two epochs. These are the first detections of water maser emission towards this source. On 2021 November 18, we detected a weak H 2 O maser at LSR =1.18 km s −1 . On 2022 January 25 we detected two features at LSR = −10.51 and LSR = 5.04 km s −1 but the 1.18 km s −1 feature had disappeared. This variability is     Strom et al. 1976). An optical outburst was detected in the late 1980's (Mauron & Thouvenot 1991) and observations by Eisloeffel et al. (1991) confirmed it showed EXor properties. The variable name V512 Per was assigned in the 71 st Name-List of Variable Stars by Kazarovets et al. (1993), who noted SVS 13 and V512 Per were the same source. A radio counterpart of the optical/near-infrared source, named VLA 4, was first detected by Rodríguez et al. (1997) and later resolved into a binary (VLA 4A and 4B; Anglada et al. 2000). Rodríguez et al. (2002) note that SVS 13 (therefore V512 Per) and VLA 4 are the same source, consistent with other studies (see, e.g., Goodrich 1986;Fujiyoshi et al. 2015). The source is also commonly known as SVS 13A (see, e.g., Plunkett et al. 2013, and references therein) and is associated with several Herbig/Haro objects (HH 7-11; e.g., Rodríguez et al. 1997;Bachiller et al. 2000). In this paper we refer to the source as V512 Per, noting that this name might be more familiar to the variable star community (e.g., Kazarovets et al. 1993;Audard et al. 2014) while SVS 13, VLA 4, or SVS 13A may be more familiar to the radio astronomy community (e.g., Rodríguez et al. 2002;Plunkett et al. 2013). Figure 3 shows the spectra obtained towards V512 Per in 2021 November. On 2021 November 18, we detected at least 6 maser features towards V512 Per (see Figure 3), with the brightest one being 20.2 Jy. Here we note that only 5 of them are shown in Table 2 (Haschick et al. 1980). H 2 O(A) is associated with V512 Per. H 2 O(B), also known as HH 7-11(B), VLA 2, SVS 13C, or MMS3, is a Class 0 source located ∼0.5 , to the southwest (Cesaroni et al. 1988;Segura-Cox et al. 2018;Chen et al. 2013;Plunkett et al. 2013), while H 2 O(C) is ∼2.5 southeast of V512 Per (Haschick et al. 1980).
To investigate which of the observed velocity components may be associated with V512 Per, we carried out a nine-point grid of observations centred on V512 Per on 2022 February 5 (with pointings separated by 20 ). The results indicate that the strong water maser features at 5-10 km s −1 are brightest at an offset position (−20 ,−20 ) rather than toward V512 Per (0 ,0 ), suggesting that these maser features do not arise mainly from V512 Per. The ∼12 km s −1 component, in contrast, is strongest towards V512 Per and is likely associated with the eruptive source ( Figure 4, see also Figure 5).

A water maser flare in H 2 O(B)
In addition to the nine-point map described above (Sect. 3.2.1), we also performed OTF mapping towards V512 Per and H 2 O(B), shown in Figure 5. As illustrated by the channel maps in Figure 5, spectral features at LSR ≤11 km s −1 peak around H 2 O(B) while spectral features at LSR >11 km s −1 peak around V512 Per. Figure 4 compares our pointed observations toward V512 Per and H 2 O(B) on 2022 February 5: the spectra show very similar profiles between 4 km s −1 and ∼10 km s −1 but the intensities are different by a factor of ∼20. This similarity suggests that our pointed observations of V512 Per, including those shown in Figure 3, have significant contributions from H 2 O(B). We estimate this contribution for our 2022 February 5 observations assuming a perfect Gaussian beam pattern with a beam size of 40 . A source at an offset of 38.7 (the angular separation between V512 Per and H 2 O(B) derived from our observations, see Table 1) will fall at the 7.5% response level of the beam, or between the 3.7-14% levels assuming a typical pointing error of 5 . Thus H 2 O(B), with a flux density of 498.7 Jy, would contribute 18.4-69.8 Jy to the spectrum observed towards V512 Per, comparable to the observed value of 21.3 Jy (Table 2).
Notably, in our pointed 2022 February 5 observations, the peak flux density of the water maser in H 2 O(B) is 498.7 Jy at LSR = 6.1 km s −1 . This is the highest flux density reported for this source to-date (c.f. Haschick et al. 1980;Lyo et al. 2014), indicative of a maser flare (see also Sect. 4.1).
In our pointing towards RNO 1B/1C, we detected water maser emission in four epochs, as shown in Figure 6. During our first observations on 2021 November 18, we detected two maser features at LSR = −28.78 km s −1 and LSR = −15.79 km s −1 , and five days later the flux densities and LSR velocities of the two maser features were nearly unchanged. The source was observed again on 2022 January 25 and February 5: in these observations, the LSR ∼ −15.8 km s −1 feature had disappeared and the blueshifted maser was weaker and had slightly shifted in velocity, to LSR ∼ −28.48 km s −1 . The 3σ upper limits for the LSR ∼ −15.8 km s −1 feature are 0.12 Jy and 0.15 Jy for the observations on 2022 January 25 and February 5, respectively. We also note that the LSR ∼ −28 km s −1 feature has the largest velocity offset with respect to the cloud among our detections, ∼10 km s −1 (see Table 2). Based on comparing our results to the literature, the water maser features detected in our survey are most likely to originate from IRAS 00338+6312 rather than RNO 1B/1C. The velocities of our detected masers are similar to those of the masers associated with IRAS 00338+6312 in the VLA observations (Fiebig 1995;Fiebig et al. 1996), and also match the velocity range of the molecular outflow (about −30 km s −1 to −5 km s −1 Snell et al. 1990;Yang et al. 1991) driven by IRAS 00338+6312 (Henning et al. 1992;Wouterloot et al. 1993;Anglada et al. 1994;Furuya et al. 2003;Bae et al. 2011). We therefore do not count the water maser emission in our RNO 1B/1C pointing as a detection towards an eruptive star, and the 3σ upper limits are given in Table B.1.

Long-term time variation
Water maser flares have been recognized in star forming regions for decades (e.g., Boboltz et al. 1998;Kramer et al. 2018), with recent observations suggesting that water maser flares can accompany ejection events associated with accretion bursts in massive and intermediate-mass stars (e.g., MacLeod et al. 2018;Brogan et al. 2018;Chen et al. 2021;Bayandina et al. 2022). Hence, one might expect such water maser flares from FUors/EXors. We therefore investigate if our targets have experienced water maser flares. Figure 7 presents long-term time series for the water masers detected in our survey, which show that these masers are quite variable in both flux density and LSR velocity. Based on data from the literature, Z CMa appears to be in a relatively active phase, with the flux density of 2.4 Jy during our observations the highest observed to date (c.f. Blitz & Lada 1979;Thum et al. 1981;Deguchi et al. 1989;Scappini et al. 1991;Palla & Prusti 1993;Moscadelli et al. 2006;Sunada et al. 2007;Bae et al. 2011;Kim et al. 2018). For HH 354 IRS, no water maser emission was detected by previous observations (Wouterloot et al. 1993;Persi et al. 1994;Sunada et al. 2007). We report the first water maser detection toward this source. Since the upper limits of previous observations are comparable to the detected flux densities (see Figure 7), we cannot conclude whether the maser was in its active or quiescent phase during our observations. For V512 Per, Figure 7 compares the velocity component in our observations that likely arises from V512 Per (see Sect. 3.2.1) to archival data that include both single-dish and interferometric measurements (Haschick et al. 1980;Claussen et al. 1996;Rodríguez et al. 2002;Furuya et al. 2003).We note that in the case of the Claussen et al. (1996) data, the results were measured from the published figures. Based on this comparison, we identify three water maser flares, in 1978, 1992, and 1998, which reached peak flux densities of ∼310 Jy, 660 Jy, and 244 Jy on 1978 February 17, 1992 November 28, and 1998 June 22, respectively. The observations spanning these dates were performed with single-dish telescopes with large beams (>1 ), so H 2 O(B) could potentially contribute to the observed flux densities (see Sect. 3.2.1). Claussen et al. (1996) note, however, that the maser features detected in their 1991-92 observations all had velocities consistent with those of H 2 O(A)/V512 Per, suggesting that this flare was associated with the eruptive star.
For H 2 O(B), we find no suggestion in the literature of this source being an eruptive variable at optical or near-infrared wavelengths, but our comparison with previous water maser observations (Figure 7; Haschick et al. 1980;Lyo et al. 2014) shows three maser flares with peak flux densities of >100 Jy, on 1975 November 30, 2012 May 28, and 2022 February 5. As for V512 Per, Figure 7 compares the velocity components in our observations that likely arise from H 2 O(B) (Sect. 3.2.1&3.3.1) with historical data. Again, the large single dish beams encompass both H 2 O(B) and V512 Per, meaning that we cannot rule out a contribution from V512 Per to the historical flares. For instance, the observations of Lyo et al. (2014) had a HPBW of 120 . As noted in Sect. 3.3.1, the water maser flare detected in our observations on 2022 February 5 is the brightest to date, with a peak flux density of 498.7 Jy.
For IRAS 00338+6312, there is similarly no suggestion in the literature of this being an eruptive source in the optical or near-infrared, but Figure 7 suggests its water maser emission was in an active phase in 1998 and 2004 (Cesaroni et al. 1988;Henning et al. 1992;Wouterloot et al. 1993;Persi et al. 1994;Fiebig Fig. 5: Channel maps of H 2 O masers in H 2 O(B) (SVS 13C) and V512 Per (SVS 13A). The contours start at 0.5 Jy, and then increase by a factor of two. The plus signs represent the positions of the two H 2 O masers (orange and green) previously detected by Haschick et al. (1980) and YSOs (purple; e.g., Plunkett et al. 2013). Based on previous observations (Plunkett et al. 2013;Podio et al. 2021), the outflow directions are indicated by red and blue arrows. The beam size is shown in the lower right corner of the last panel. The colour bar represents the flux density in units of Jy.  (Szabó et al. 2023(Szabó et al. ). 1995Codella et al. 1995;Furuya et al. 2003;Sunada et al. 2007;Bae et al. 2011), but relatively quiescent during our observations. The highest flux density reached was ∼31 Jy on 1998 January 5 (Furuya et al. 2003).
Periodic variations have been reported in some velocity components of the 22.2 GHz H 2 O (and the 6.7 GHz Class II CH 3 OH) masers associated with the intermediate-mass YSO G107.298+5.639, and cyclic accretion instabilities have been invoked to explain this peculiar behavior (Szymczak et al. 2016). Low-mass stars like FUors and EXors might also experience cyclic accretion events, but we do not find evidence for periodic variations in Figure 7.

Scarcity of water masers in selected eruptive systems
Our water maser detection rate of 6% in FUors and EXors is perhaps surprising in light of the close connection between water maser emission and mass accretion and ejection in protostars (see Sect. 1). In this section, we consider possible explanations for the low detection rate.
First, the low detection rate could be caused by an evolutionary effect. Previous observations indicate that the water maser detection rate decreases from Class 0 to Class II objects (e.g., Furuya et al. 2001 Class I and Class II objects (see Tables 1 and B.1), one would expect a lower detection rate compared to Class 0 objects. Furthermore, our detection rate is comparable to that (6.3%) for Class I objects in Furuya et al. (2001). We do not detect any water masers toward Class II objects, which further supports the evolutionary trend proposed by Furuya et al. (2001). Second, water masers have relatively low luminosities in low-mass star formation regions. Statistical studies have shown that the maser luminosities are correlated with bolometric luminosities (e.g., Figure 16 in Urquhart et al. 2011). This suggests lower maser luminosities in low-mass star formation regions, so lower flux densities would be expected. This could contribute to our low detection rate toward low-mass eruptive stars. This is supported by previous water maser surveys toward the Serpens South and Orion molecular clouds (Kang et al. 2013;Ortiz-León et al. 2021), which give detection rates of 2% for low-mass protostars.
Third, water masers show rapid time variations. The time variability of water masers is evident in our study (see also Figures 2, 4 and 7). Water masers can be in a quiescent phase for ∼5 years (Claussen et al. 1996), meaning that maser emission would not be detected during that time even for sources known to be associated with water masers. This is consistent with the fact that several water masers reported by previous studies are not detected in our observations (see Table B.1). It is possible that non-detection of water masers is due to their inactive state. Indeed, including historical detections, the detection rate of water masers in eruptive stars in our sample is ∼15% (excluding the unclassified Gaia alerts), which is higher than our survey detection rate of 6%, suggesting that previously detected water masers were in an inactive phase during our observations.

Conclusions
In this paper, we presented the results of the first dedicated water maser survey towards FUors and EXors, two classes of low-mass young eruptive stars. We detected H 2 O masers toward five objects, of which three are young eruptive stars: Z CMa (FUor; Class I), HH 354 IRS (FUor; Class 0/I), V512 Per (EXor; Class I), IRAS 00338+6312 (Class 0) and H 2 O(B) (Class 0). Our detection is the first report of water maser emission in HH 354 IRS. Our observations reveal the highest peak flux density yet reported towards H 2 O(B) (498.7 Jy), indicative of a recent H 2 O maser flare. Overall, our observations result in a detection rate of ∼6% for young eruptive stars. Analysis of the longterm time series of the water masers suggests that V512 Per and H 2 O(B) have experienced multiple water maser flares.
Despite the low detection rate, our observations have confirmed the presence of 22.2 GHz water maser emission in FUors and EXors, meaning that follow-up radio interferometric observations can be used to probe the environments of eruptive stars on small scales (see, e.g., Haschick et al. 1980;Rodríguez et al. 2002). If water masers are in general weak in FUors/EXors (Sect. 4.2), deeper observations would also be expected to find more of them. Expanding on optical and near-infrared knowledge of FUors/EXors with more radio observations, especially future VLBI measurements, will be crucial to better understand the underlying physics (e.g., mass accretion and ejection) of such peculiar objects, and eventually the formation of Sun-like stars.