H i imaging of dwarf star-forming galaxies:
Masses, morphologies and gas deficiencies

The H$\alpha$ and $R$-band observations were carried out using the 1.3-m Devasthal Fast Optical Telescope (DFOT) and 2-m IUCAA Girawali Observatory (IGO) telescope in the years 2011 - 2013. The details of the optical observations and data reduction are provided in Jaiswal & Omar (2016). The H$\alpha$ images have total exposure time in the range $30-235$ minutes while the R-band images have exposures in the range $15-50$ minutes. The point spread function has FWHM values in the range $1.2^{\prime\prime}-2.6^{\prime\prime}$. The CCD images were processed following the standard tasks of IRAF (Image Reduction and Analysis facility) software developed by National Optical Astronomy Observatory. The H$\alpha$ images have surface brightness sensitivities in the range $0.4-1.8~{}\times 10^{-16}~{}\rm erg~{}s^{-1}~{}cm^{-2}~{}arcsec^{-2}$ for 3$\sigma$ detection. These optical images are used here for overlaying H i contours.

The H i observations of 13 galaxies were carried out using the GMRT (Swarup et al., 1991). The GMRT is a Y-shaped interferometric array of thirty fully steerable parabolic dishes of 45-m diameter each. The half power beam width of the GMRT antennas is $\sim 24^{\prime}$ at 1.4 GHz. The GMRT has a mix of both short and long baselines, making it sensitive to diffuse emission of extent as much as $7^{\prime}$ while having a maximum possible resolution of $\sim 2^{\prime\prime}$ at 1.4 GHz. The observations of the galaxies were carried out interspersed with the observations of a suitable nearby phase calibrator every 45 minutes and observations of at least one of the primary flux calibrators 3C48, 3C147, and 3C286. The phase calibrators were selected from the Very Large Array list of calibrators within an angular distance of 10^∘ to the target galaxy. The bandpass calibration was performed using the primary flux calibrators. A summary of the observational parameters are given in Table 2. The data were taken in right and left circular polarization sensitive for total intensity. The digital correlator back-end was configured to use either 4.2 or 8 MHz bandwidth in 256 or 512 spectral channels. The centre frequency of the band-width was set corresponding to the redshifted H i emission line frequency from the galaxy.

The data were reduced following the standard calibration and imaging methods using the Astronomical Image Processing System (AIPS: Greisen, 2003) developed by the National Radio Astronomical Observatory. The data were calibrated for amplitude, phase and frequency response for all the antennas and separately for each of the two polarizations. The data were also self-calibrated in phase and amplitude using the continuum emission formed in a 100-channel averaged dataset. The resulting complex gain corrections for the antennas were applied to all the frequency channels. The continuum emission common to all the channels was subtracted from the line-data using the ‘UVLIN’ task in AIPS. The self-calibrated and continuum subtracted line-data were used to make the image cubes at different resolutions by selecting appropriate spatial frequency range, taper and robustness parameter as implemented in the AIPS task ‘IMAGR’. The image cubes were generated and de-convolved using the CLEAN algorithm as implemented in the AIPS task IMAGR. The Doppler frequency shift corrections arising due to the orbital and spin motions of the Earth relative to the galaxy were applied on the image cubes to bring all the velocities in the Helio-centric frame of reference. The frequency was converted to velocity following the optical definition. The images were corrected for the primary beam of the GMRT antenna using the AIPS task PBCOR. The channel images were further analysed for estimating various parameters as described below.

The moment maps at different resolutions provide complementary information, such as the low-resolution images provide large-scale dynamics, global extent of H i, effect of the environment and dark matter halo properties, while the high-resolution images trace the local phenomena like star formation and its feedback, and finer details of the H i morphology near the central regions of the galaxies. The low-resolution H i column density contour image plotted on the grey-scale optical $R$-band image shows the overall H i morphology of the galaxy with respect to its optical morphology. These images are sensitive to detect tidal interaction features and nearby companion, if any. The intermediate resolution H i column density contours plotted over the grey-scale optical $R$-band image provides visual correlation with the stellar disk and detection of lopsidedness features. The intermediate resolution moment-1 color maps show the velocity field of the galaxies. In most of the cases, the velocity field is found to be slightly irregular or disturbed. It was not possible to obtain spatially resolved reliable rotation curves of these galaxies due to lack of adequate SNR at high angular resolution. The high resolution H i column density contours plotted on grey-scale H$\alpha$ image for the galaxies provides correlation of H i with the star forming regions (e.g. van Zee et al., 1998). The star-forming regions in these dwarf galaxies are mostly located in the central regions. In some cases, asymmetries in the H i distribution with respect to the optical image of the galaxy were seen.

The estimated H i parameters are given in Table 4. The columns in this table stand for the galaxy name, H i flux integral, corrected line-widths at 20% and 50% levels, optical inclination angle, galaxy rotation velocity, galaxy systemic velocity, H i mass, H i disk radius and H i mass to blue light ratio, respectively. The H i mass to blue light ratio is a scale-free H i parameter as it is independent of the distance to the galaxy. The blue luminosities of the galaxies were estimated using the $B$-band magnitudes from literature (see, Table 5) and the extinction corrections from Jaiswal & Omar (2016) in the relation $L_{B}/L_{B,\odot}=10^{0.4(M_{B,\odot}-M_{B})}$ with $M_{B,\odot}=5.515\pm 0.020$ mag (Ramírez et al., 2012; Pecaut & Mamajek, 2013). The comparison of the H i flux integral estimated using the GMRT observation and that from the single-dish measurements for the sample galaxies available in the literature is shown in Figure 2. In many cases, the two measurements are different even within uncertainties. In the plot of logarithm of H i flux integral ratios, majority of galaxies show deviation $<$0.1 dex. A few galaxies show deviations up to 0.3 dex. Considering relatively large errors in the single-dish measurements, this comparison indicates that the GMRT measurements are fairly accurate and do not suffer from the interferometric flux-loss for the sizes of the galaxies studies here. The H i mass surface-density profiles of the sample galaxies are shown in Figure 3. The average peak surface density of these galaxies is $\sim 2.5~{}M_{\odot}~{}{\rm pc}^{-2}$. The optical disk radius at 25 mag arcsec^-2 surface brightness in the $B$-band image of the sample galaxies are given table 5 from the HyperLEDA database (Paturel et al., 1991; Paturel et al., 1997). The error-weighted average value of the H i diameter to optical diameter ratio ($D_{\rm HI}/D_{25}$) is estimated to be $1.9\pm 0.1$, which is higher than those estimated for spiral galaxies in field and group environment (Broeils & Rhee, 1997; Verheijen & Sancisi, 2001; Omar & Dwarakanath, 2005a). It is worth to point that due to strong tidal features in UGCA 116, the H i diameter of UGCA 116 is poorly constrained. The H i radius for UGCA 116 is estimated by assuming a face-on system. The H i radius is taken along the tidal tail, hence the ratio of its H i diameter to its optical diameter is significantly different from the average value.

The H i morphologies indicate that the majority of these galaxies have features indicative of tidal interaction such as lopsidedness, tails and bridges, plumes and irregular velocity field. In many cases, the centre of H i emission is offset from that of the H$\alpha$ line or optical continuum emission. Often additional H i components with no optical counterpart are found in the vicinity of the galaxy. As the dynamical ages of such tidal features are typically 100 Myr, these features often trace very recent tidal encounters which may be responsible for the young and massive star-formation in the galaxies. The following notes summarize main features seen in the H i images of the individual galaxies:

The H i morphologies of all the galaxies in the sample were visually checked for signatures of tidal interactions and presence of H i clouds in the vicinity. A summary of the prominent optical and radio morphological features seen in the WR galaxies in our sample is provided in Table 6. The H$\alpha$ and optical morphology column in this table is taken from Jaiswal & Omar (2016). The interaction probability in this table is inferred based on various morphological features. The presence of multiple nuclei, arcs and tidal tails are generally considered as a signature of recent tidal encounter (e.g. Beck & Kovo, 1999; López-Sánchez et al., 2004; Matsui et al., 2012; Adamo et al., 2012). Galaxies showing H i or optical tidal tails, irregular velocity field, presence of H i clouds in the vicinity or distinct multiple nuclei and a disturbed optical envelope are termed as highly probable. Galaxies showing lopsidedness or asymmetric light distribution along with any other feature such as probable multiple nuclei or arcs/plumes are termed as moderately probable. Galaxies without any significant disturbed optical or H i morphology are labeled with low interaction probability. It should be noted that the galaxies termed as having low probability may show some interaction features such as bar, mild lopsidedness, or nuclear star formation. However, due to the absence of any strong feature of recent tidal interaction, we preferred to label such objects with low interaction probability.

In the present sample, we find that three galaxies show lopsided or disturbed radio continuum emission based on analyses presented in Jaiswal & Omar (2016). Based on the present study, six galaxies show lopsidedness in H i and nine galaxies are found to have disturbed H i morphologies and velocity fields. The lopsidedness in galaxies in general has been studied previously and is believed to be caused by tidal interactions (e.g. Jog, 1997; Zaritsky & Rix, 1997; Angiras et al., 2006). Majority of dwarf galaxies in the present sample are showing intense nuclear star-burst consistent with the previous results (e.g. Strickland & Stevens, 1999; Adamo et al., 2011). It has also been predicted via N-body simulations (e.g. Hernquist & Mihos, 1995) that the tidal interaction and mergers between galaxies can channel gas towards the centre of galaxy as a result of loss of angular momentum. This gas can give rise to nuclear starburst in galaxies (Mihos & Hernquist, 1994). The optical and radio morphological features seen in our study are also highly suggestive of prevalence of tidal interactions and mergers in the WR galaxies. A total of 12 galaxies are inferred here to be tidally interacting based on optical and H i morphologies. The remaining one galaxies, namely SBS 1222+614, is not detected in H i line imaging. However, the detected galaxy MCG +10-18-044 in the vicinity of SBS 1222+614 is found to have some interaction features. These results are in agreement with the previous studies on the dwarf WR galaxies, where tidal interactions were inferred (Méndez & Esteban, 1999; López-Sánchez & Esteban, 2008; López-Sánchez, 2010). The main cause of starburst in several dwarf and blue compact galaxies appear to be tidal interactions with a nearby companion or low mass H i cloud without any known optical companion. The optical companion may be an ultra-faint or low surface brightness dwarf galaxy not detected in present optical images. The H i clouds may also be anomalous velocity H i clouds falling onto galaxy as a result of past merger or tidal interaction with a galaxy. Alternatively, some of these clouds may be a result of accretion of gas from the inter-galactic medium. It is not possible to favour or reject a possibility based on the present observations. Combined with previous studies on the dwarf WR galaxies, our analysis suggests that overall more than three-fourths of WR galaxies show obvious signatures of recent tidal encounters and mergers.

Recently Teich et al. (2016) and McNichols et al. (2016) have performed a “Survey of H i in Extremely Low-mass Dwarfs (SHIELD)” having 12 objects observed through the Karl G. Jansky Very Large Array (VLA) at a variety of spatial and spectral resolutions. They also found the offsets between H i and star formation peaks in galaxies with the lowest total H i masses. SHIELD galaxies’ dynamics are found to be equally supported by rotational and pressure contributions. However, in the present sample, most of the galaxies are showing disturbed velocity field.

The main mass components of a late-type galaxy are gas, dust, stars and dark matter. The total mass of gas ($M_{gas}$), dust ($M_{dust}$) and stars ($M_{stars}$) is jointly termed as the luminous mass or baryonic mass. The stellar mass contribution to the baryonic mass generally dominates over the gas mass in bright spiral galaxies. However, in dwarf star-forming galaxies the gas mass can be a large fraction of the stellar mass. The dust mass remains only a few percent of the gas mass (e.g. Draine et al., 2007; Clemens et al., 2013) and therefore can be neglected while estimating the baryonic mass. An estimate of the dark matter mass ($M_{dark}$) can be done by subtracting the baryonic mass ($M_{bary}$) from the total dynamical mass ($M_{dyn}$), which is estimated here within the H i disk radius of the galaxy. The masses of different components of a galaxy constraint the galaxy properties and therefore provide an indication of the galaxy formation and evolution (e.g. López-Sánchez, 2010).

The estimations of the gas, dust and stellar masses require analyses of different radiations coming from these components. The gas mass can be calculated by adding the neutral atomic hydrogen (H i) mass, molecular hydrogen (H₂) mass and helium (He) mass, neglecting very small contributions from the metals in the gaseous phase and the ionized gas (Rémy-Ruyer et al., 2014). The H i mass in the galaxies in the present sample are estimated using the GMRT H i flux integral and are given in Table 4. The H₂ gas mass is normally measured using the CO emission lines. The amount of H₂ gas in dwarf star-forming galaxies can be substantial. However, no CO observations are available for the galaxies in our sample and hence we could not include H₂ mass in the present analysis. The atomic gas masses (see, Table 7) are estimated as $M_{\rm H\,I}+M_{\rm He}=1.4M_{\rm H\,I}$ following Bradford et al. (2015).

The stellar mass of a galaxy can be estimated by multiplying its optical or near-infrared (NIR) luminosity with a value of stellar mass-to-light (M/L) ratio known for different wavebands. The mass-to-light ratio of stellar population varies with age and metallicity (Schombert & Rakos, 2009; Schombert & McGaugh, 2014). It is therefore an approximation to assume a single, constant mass-to-light ratio for estimating the stellar mass in a galaxy as the galaxy is likely to have stars of different ages and metallicity. A slightly sophisticated approach for estimating the stellar mass is to use the color-dependent mass-to-light ratio derived using a stellar population synthesis model. The variation of mass-to-light ratio with optical color is found to be minimum in the NIR bands in comparison to the optical wavebands (e.g. McGaugh & Schombert, 2014). Taylor et al. (2011) have found that the relation between $(g-i)$ SDSS color and mass-to-light ratio in the $i$-band is very tight for a sample of galaxies. Therefore, we also used $(g-i)$ color and the $i$-band mass-to-light ratio relation for the galaxies in the present sample to estimate stellar mas following the Bell et al. (2003, see Appendix A) relation. The scatter in this correlation is nearly 0.1 dex. Note that Bell et al. (2003) relation to estimate the stellar mass uses Salpeter (1955) initial mass function (IMF) modified with Bell & de Jong (2001) procedure, in order to provide the maximum possible stellar M/L ratio (at a given color) which can truly constrain the galaxy rotation velocity, while the normal Salpeter (1955) IMF overpredicts the galaxy rotation velocity through higher number of low-mass faint stars and thereby larger value of the stellar M/L ratio. The $(g-i)$ colors and $i$-band apparent magnitudes (see Table 5) for 10 galaxies are taken from the SDSS catalogue data release-7 (Abazajian et al., 2009). The $g$ and $i$ band magnitudes for the remaining three galaxies (MRK 996, UGCA 130 and I SZ 59), which do not have SDSS data, are extrapolated from the Johnson $B$ and Cousins $R$ band magnitudes (Gil de Paz et al., 2003) with the help of Lupton transformation relations (http://www.sdss3.org/dr8/algorithms/sdssUBVRITransform.php#Lupton2005). The $i$-band luminosities are estimated in a similar manner as the $B$-band luminosities, considering $M_{i,\odot}=4.57\pm 0.03$ mag (http://www.sdss3.org/dr8/algorithms/ugrizVegaSun.php). The masses of different mass components of the galaxies are given in Table 7. It can be seen from this table that the gas masses of some galaxies are greater than their stellar masses. The errors in the masses are estimated using error propagation of the quantities in the formula of corresponding mass given in the Appendix A.

We plotted the baryonic mass $(M_{stars}+M_{\rm H\,I}+M_{\rm He})$ against the dynamical mass and carried out a linear fit in Figure 4. We applied the condition that the baryonic mass tends to zero at the zero dynamical mass, i.e. the intercept of the linear fit on the $y$-axis is fixed to zero. This plot indicates that the baryonic mass and the dynamical mass are well correlated with the correlation coefficient $r=0.89$, which in turn shows a little or devoid of dark matter within the H i disk radius of these systems. We also find that the H i disk diameter up to 1 $M_{\odot}~{}{\rm pc}^{-2}$ H i mass surface density is larger than the stellar disk diameter for the galaxies in the sample, consistent with previous studies. We also noticed that the gas fraction, given as the ratio of the gas mass $(M_{\rm H\,I}+M_{\rm He})$ and the baryonic mass are directly correlated with their $B$-band absolute magnitude (see Figure 5), which is related to the young stellar population in the galaxies. The error-weighted average gas fraction is found to be $0.24\pm 0.01$, which is in agreement with the lower limit of $0.15$ in Bradford et al. (2015) study of non-isolated low-mass galaxies suggesting that the low gas fraction in these systems are caused by the environmental effects (see Section 5.4). The error-weighted average baryon fraction ($(M_{stars}+M_{\rm H\,I}+M_{\rm He})/M_{dyn}=0.18\pm 0.01$) for our sample is also in agreement with Bradford et al. (2015) results.

The H i-stellar mass relation is well known tool to provide an indication of the history of gas accretion in conjunction with star formation (e.g. Huang et al., 2012; Maddox et al., 2015; McQuinn et al., 2015; Parkash et al., 2018). The H i mass is found to be directly correlated with the stellar mass of galaxies in the sample and the H i-to-stellar mass fraction ($f_{\rm H\,I}=\frac{M_{\rm H\,I}}{M_{stars}}$) is anti-correlated with the stellar mass (figure 6). These correlations are consistent with the previous studies (e.g. Wang et al., 2017; Parkash et al., 2018). The average H i-to-stellar mass fraction for the sample galaxies is found to be $0.14\pm 0.01$.

The SFR-stellar mass relation for star-forming galaxies is referred as the main sequence of galaxies (Noeske et al., 2007; Speagle et al., 2014; Kurczynski et al., 2016). This relation suggests possible range of SFRs at a given stellar mass for a particular redshift and can be used to study the star formation history of galaxies. We have compared our sample with a resembling sample, both in mass and redshift, of 56 LSB galaxies from McGaugh et al. (2017) in terms of the SFR-stellar mass relation. Figure 7 shows the SFR vs stellar mass plot for our sample galaxies along with the SFR-stellar mass relation obtained from McGaugh et al. (2017). Our sample galaxies are found to have elevated SFRs compared to the reference sample. This result is consistent with the expectation as we have selected the galaxies from the catalogs of WR galaxies.

The environment of a galaxy plays an important role in deciding the gas content and the morphology of a galaxy (e.g. Fasano et al., 2000; Goto et al., 2003). It is found observationally that the galaxies located near the cluster centres have on average less H i content in comparison to those at the outskirts of the clusters or field galaxies of the same morphological type and size (Giovanelli & Haynes, 1985; Haynes & Giovanelli, 1986; Solanes et al., 2001; Gavazzi et al., 2006). The deficiency of H i in the galaxies within the cluster environment can be understood using several gas-removal mechanisms, for example tidal interactions (Mihos et al., 2005) and ram-pressure stripping (Gunn & Gott, 1972). The process of gas removal via tidal interactions depends upon galaxy density and relative velocity of the encounters. The tidal interactions for the galaxies moving slow in a high galaxy density region (e.g., in a group environment) will be more effective in removing gas from the galaxies. The gas removal via ram-pressure depends on density of matter in the inter-galactic medium (IGM) and velocity with which the galaxy is moving in the medium in the sense that at higher density and velocity (e.g., in a cluster environment), the process will be more effective. These processes affect the H i morphology in the form of H i tails and bridges, displaced H i from the disk, lopsided and truncated H i disks (e.g. Chung et al., 2007; Hibbard et al., 2001; Dénes et al., 2016). The removed H i content of a galaxy affected by the environment through the gas removal processes can be quantified as the H i deficiency parameter (Giovanelli & Haynes, 1983): $DEF_{\rm HI}={\rm log}(M_{\rm HI,exp})-{\rm log}(M_{\rm HI,obs})$, where $M_{\rm HI,obs}$ is the observed H i mass of the galaxy and $M_{\rm HI,exp}$ is the expected H i mass of the galaxy obtained using some H i-optical scaling relation for a sample of isolated galaxies. We have used the convention that the galaxies with $DEF_{\rm HI}>0.3$ are considered to be H i-deficient. Here, $DEF_{\rm HI}=0.3$ indicates two times less H i than the isolated galaxy of the same morphological type and size in the field environment. There are various H i-optical scaling relation available for the estimation of expected H i mass for a galaxy of a particular morphological type (Haynes & Giovanelli, 1984; Chamaraux et al., 1986; Solanes et al., 1996; Catinella et al., 2010; Toribio et al., 2011; Dénes et al., 2014). It is suggested (e.g. Haynes & Giovanelli, 1984) to use the distance-independent quantities like $M_{\rm HI}/L$ and $M_{\rm HI}/D^{2}$, where $L$ is luminosity and $D$ is diameter in some optical waveband, in order to obtain H i deficiency in a galaxy. The diameters are less affected by the extinction and its uncertainty in comparison to the luminosities. It has been shown Haynes & Giovanelli (1984) that $M_{\rm HI}/D^{2}$ is a better parameter to represent the H i deficiency in comparison to $M_{\rm HI}/L$ by measuring the uncertainty in these quantities for individual galaxies. They also noted that $M_{\rm HI}/D^{2}$, in contrast to $M_{\rm HI}/L$, is nearly independent of the morphological type of the galaxy. This is because $M_{\rm HI}$ and $D$ are the disk properties only, while $L$ includes the contribution due to the (gas-free) spheroidal bulge whose contribution to the optical luminosity varies with the morphological type of the galaxy. We therefore use H i mass and optical diameter scaling relation to estimate the expected H i mass of the galaxies and the H i deficiency parameter. Based on the above arguments and the availability of the photometric data, we have therefore used the scaling relation between the H i mass and the $B$-band isophotal diameter at 25 mag arcsec^-2 surface brightness derived using an un-biased sample of galaxies given by Dénes et al. (2014) for 844 galaxies having H i masses between $10^{8}$–$10^{10.5}$ $M_{\odot}$.

Figure 8 shows the plot of H i mass versus $B$-band isophotal diameter. The solid line indicates the H i mass and $B$-band diameter scaling relation from Dénes et al. (2014) and the dashed lines indicate $1\sigma$ scatter of $\pm 0.3$. It can be seen that most of the galaxies in our sample are H i deficient. Only UGCA 116 is found to have its normal H i content. The effect of environment on the deficiency parameter is investigated by plotting the galaxy density around the sample galaxies versus the deficiency parameter (Figure 9). The galaxy density is defined here as the number of galaxies within the co-moving volume having 1 Mpc projected radius and $\pm 500$ km s^-1 radial velocity range around the target galaxy. The environmental galaxy densities around each of the sample galaxies are taken directly from the NED tool(https://ned.ipac.caltech.edu/forms/denv.html) and are given in Table 5. The NED-measured environment uses the spectroscopic redshift information of the sources given in the literature and optical/NIR/radio line survey catalogues, for example, SDSS catalogues (York et al., 2000), Third Reference Catalogue of bright galaxies (RC3: de Vaucouleurs et al., 1991), The Updated Zwicky Catalog (UZC: Falco et al., 1999), NOAO Fundamental Plane Survey (Smith et al., 2004), The DEEP2 Galaxy Redshift Survey (DEEP2: Davis et al., 2003), The CFA Redshift Survey (Huchra et al., 1999), The 6dF Galaxy Survey (Jones et al., 2009), The Southern Sky Redshift Survey (da Costa et al., 1998), The Smithsonian Hectospec Lensing Survey (SHELS: Geller et al., 2016), 2MASS Redshift Survey (Huchra et al., 2012), Neutral Hydrogen Surveys (Schneider et al., 1992; Theureau et al., 1998; Barnes et al., 2001; Paturel et al., 2003; Tully et al., 2008; Haynes et al., 2011). As our sample galaxies are nearby systems, the redshifts of most of their environment sources are known accurately. Estimated galaxy densities from NED are only a lower limit due to a possibility of unidentified members surrounding the galaxies in the sample. Our aim here was to assess the environment, which appears to be similar to groups of galaxies for almost all WR galaxies studied here. The number of group members within $\sim$1 Mpc radius for the WR galaxies studied here is 5–31. The galaxy densities are similar to typical group environments such as those in the Eridanus groups of galaxies (Omar & Dwarakanath, 2005b) and Sengupta & Balasubramanyam (2006). The primary cause of H i deficiency in these groups are identified as tidal interactions. It is noticed that UGCA 116 having no deficiency is lying in the lowest galaxy density ($\sim$1.2 Mpc^-3) region and all the galaxies with significant deficiency are residing in a relatively higher galaxy density ($<8$ Mpc^-3) regions. This trend is similar to that noticed in Omar & Dwarakanath (2005b).

The variation of the H i mass with the H i diameter is plotted in Figure 10. The H i diameter is estimated at the limiting H i mass surface density as $1~{}M_{\odot}~{}{\rm pc}^{-2}$. The H i masses of the galaxies are tightly correlated with their H i disk diameter with a correlation coefficient of $1$ and a scatter of $0.08$. The linear-fit to the galaxies in our sample has the form of ${\rm log}M_{\rm HI}=(2.05\pm 0.12)~{}{\rm log}D_{\rm HI}+(6.31\pm 0.10)$, in good agreement with the previous studies (Haynes & Giovanelli, 1984; Broeils & Rhee, 1997; Verheijen & Sancisi, 2001; Omar & Dwarakanath, 2005a). It is worth to point out that the late-type galaxies in the Eridanus group studied by Omar & Dwarakanath (2005b) show H i deficiency, however, average as well as peak H i mass surface densities of the majority of the Eridanus galaxies are much higher and comparable to the massive spirals in the field environments. The dwarf galaxies studied here, on the other hand, have low H i mass surface density throughout the H i disk although the H i disk extends far beyond the optical disk. The H i mass surface densities of the galaxies studied here are however comparable to those in dwarf irregular galaxies (Begum et al., 2008). A tight relation between H i mass and H i diameter indicates that average H i mass density is fairly constant in all the galaxy types.

The H i content of a galaxy is an important parameter to study galaxy evolution (Haynes & Giovanelli, 1984). For example, the gas-rich late-type galaxies in high-density regions such as in galaxy clusters can be transformed into the gas-poor galaxies over time, under the effect of several gas removal mechanisms, such as mergers and tidal interactions (Mihos et al., 2005), ram pressure stripping (Gunn & Gott, 1972), turbulent or viscous stripping (Nulsen, 1982), thermal evaporation (Cowie & Songaila, 1977), starvation (Larson et al., 1980) and harassment (Moore et al., 1996). However, in low density environment such as loose groups, the gas removal is generally not very intense as compared to that in the galaxy clusters. The main mechanisms for the gas removal in groups is identified as tidal interaction (e.g. Kern et al., 2008; Rasmussen et al., 2008) or a combination of ram-pressure stripping and tidal interaction (e.g. Davis et al., 1997; Mayer et al., 2006; Rasmussen et al., 2012). As the galaxies in our sample are in a group-like environment, it is therefore expected that the tidal interaction mechanism is responsible for the H i deficiency of the sample galaxies. It is also consistent with the H i morphologies discussed in Section 5.1.