5 Results

The $\rm H_\alpha$ EW of galaxies is known to increase systematically along the Hubble sequence, from virtually zero for the early types (E-S0) to several hundred Å for the latest types (Kennicutt 1998). A weak trend is confirmed when data limited to the Virgo spiral galaxies included in this work are used, as shown in Fig. 5. However the scatter in each of the morphological type bins is as much as an order of magnitude, even though the scatter is somewhat reduced when gas deficient galaxies are excluded. The Hubble type alone does not account for the star formation properties of galaxies in this cluster. To shed light on other possible dependences we will analyze how the SFR varies as a function of the projected clustercentric distance (Sect. 5.1), of the luminosity (Sect. 5.2) and of the gaseous content of galaxies (Sect. 5.3).

Figure 5: The distribution of $\rm H_\alpha$ EW of spiral galaxies in the Virgo cluster as a function of Hubble type. Filled dots represent galaxies with normal gas content ( $Def_{{\rm gas}}<0.4$), open symbols are gas deficient objects. To avoid superposition of points, galaxies in each type bin are separated by a small random quantity. Crosses represent averages (including only the non gas-deficient galaxies) in each morphological type bin.

5.1 The clustercentric dependence of the SFR

Lewis et al. (2002) analyzed the dependence of the galaxy SFR on the projected distance from clusters in the 2dF survey. Their volume limited samples comprise galaxies of all morphological types with 0.05<z<0.1, brighter than M_b<-19. They showed with high statistical significance that the median SFR of galaxies decreases with decreasing projected distance from clusters.

It would be interesting to compare these intermediate distance clusters with the 3 local clusters analyzed in this work, however a direct comparison cannot be carried out because data of early-type galaxies are not in our possession. The dependence of the $\rm H_\alpha$ EW on the clustercentric distance in units of virial radii can be analyzed only for the late-types galaxies, bearing in mind that our completeness is 60%. We compute $R_{{\rm virial}}=0.002\sigma_r h^{-1}$ (Girardi et al. 1998) for the 3 clusters assuming $\sigma_r$ = 775, 840, 925  $\rm km ~s^{-1}$ for Virgo, A1367 and Coma respectively.

The combined Coma and A1367 clusters (with M_b<-19) are shown in Fig. 6 enbedded in the Coma supercluster that we trace out to large clustercentric radial distances. We find a significant inner decrease only of the 25th percentile of the $\rm H_\alpha$ EW distribution. Both the median and the 75th percentile instead increase inwards. We find it unlikely that the $\rm H_\alpha$ EW distribution is biased toward high values due to incompleteness, since for the Coma+A1367 clusters our survey covers 75% of the sample. These clusters are inhabited by strong $\rm H_\alpha$ emitters to which the attention has been drawn by several authors. These include the "blue galaxies in the Coma cluster'' of Bothun & Dressler (1986) and the blue galaxy sample observed with ISO by Contursi et al. (2001). Many (13) galaxies with $\rm H_\alpha$ EW in excess of 50 Å  are found both in the inner regions ( $R/R_{{\rm virial}}<0.5$) and at intermediate distances ( $0.5<R/R_{{\rm virial}}<1.5$) from the observed clusters. Noticeably these galaxies are near the faint limit of our survey ( -19.5<M_b<-19 mag).

For the Virgo cluster we separate the bright sample (M_p<-19), with a luminosity cutoff and $\rm H_\alpha$ completeness similar to the Coma supercluster (75%), from the total sample (M_p<-15) and we show the two radial dependences separately in Fig. 7. The bright sample shows an inner decrease of the SFR. For the total sample this pattern no longer holds true. The Virgo cluster contains 24 galaxies with $\rm H_\alpha$ EW in excess of 50 Å  (11 are BCDs), the majority (14/24 objects) being fainter than -17.1 mag.

Because of their low optical luminosity the strong $\rm H_\alpha$ emitters belonging to Virgo would have all escaped detection in the 2dF survey. We conclude that, beside morphology segregation, the three local clusters analyzed in this work do not show a clear radial trend of the SFR distribution. The presence of the radial trend depends purely on a luminosity cutoff, which varies cluster to cluster between -17 and -19 mag. While spiral galaxies brighter than this cutoff luminosity have lower than average SFR at the cluster centers, galaxies fainter than this limit have SFR independent from the clustercentric projected distances. This is consistent with the idea that infall of small galaxies is occurring onto rich clusters at the present cosmological epoch.

Figure 6: The distribution of $\rm H_\alpha$ EW as a function of (projected) clustercentric radius from the Coma and A1367 clusters (Mb<-19). The top and bottom lines represent the 75th and the 25th percentile of the EW distribution, while the central line is the median of the distribution.

Figure 7: The distribution of $\rm H_\alpha$ EW as a function of (projected) clustercentric radius from the Virgo cluster. The top and bottom lines represent the 75th and the 25th percentile of the EW distribution, while the central line is the median of the distribution. The top panel shows the Virgo galaxies brighter than Mp=-19, while the bottom panel includes all galaxies surveyed in $\rm H_\alpha$ (Mb<-15).

5.2 The SFR in the Coma supercluster

Figure 8: The relation between the birthrate parameter and the NIR luminosity (mass) for the Coma supercluster galaxies. Galaxies in the Coma+A1367 clusters are represented with empty symbols, filled symbols are non-cluster galaxies. The dotted line represents the expected b as a function of $L_{\rm H}$ in the closed-box model of Eq. (5). The dashed line represents the observational bias affecting the Coma galaxies due to their selection in the B band.

Since, as concluded in the previous section, the present star formation rate of galaxies near the center of the studied clusters is a luminosity sensitive parameter, it is compelling to proceed to a systematic investigation of the luminosity dependence of the star formation properties. To this aim it is adequate to analyze the luminosity dependence of the birthrate parameter b (see Sect. 4.1). The most appropriate luminosity indicator, which we will adopt hereafter, is the NIR (H band) luminosity. This parameter traces at best the dynamical mass (within the optical disk) of spiral galaxies, as concluded by Gavazzi et al. (1996c), who found ${\rm Log} M_{{\rm dyn}}= {\rm Log} L_{\rm H} + 0.66$.

The dependence of the b parameter on $L_{\rm H}$, given in Fig. 8, shows that the star formation history of spiral galaxies in the Coma supercluster region is in almost inverse proportionality with the system luminosity (mass). The most massive spirals ( ${\rm Log} L_{\rm H} \sim 11.5 L_{{\rm H}_\odot} \sim 12.3~M_\odot$) have their b parameter as much as 100 times lower than less luminous (giant) galaxies ( ${\rm Log} L_{\rm H} \sim 10 L_{{\rm H}_\odot} \sim 10.8~M_\odot$). This confirms previous claims that the current SFR, as derived from the  $\rm H_\alpha$ EW, anti-correlates with the system mass (Gavazzi et al. 1998). Furthermore Fig. 8 shows that there is not a significant difference between the SFH of galaxies in the rich Coma+A1367 clusters and of relatively isolated objects in the same supercluster, in agreement with Gavazzi et al. (1998).

Both results are however biased by selection effects. The Coma supercluster galaxies were selected optically in the blue (photographic) band ( $m_p \leq ~15.7$). The selected targets were observed "a posteriori'' in $\rm H_\alpha$ and in the NIR, therefore at any given $L_{\rm H}$ only galaxies bluer than a certain threshold are sampled. In other words the B selection biases against faint-red galaxies, according to the relation between B-H and the infrared luminosity represented by Eq. (9) (see Scodeggio et al. 2002). This, combined with the fact that b correlates with the B-H color (see Eq. (10)), introduces a selection effect in the b vs. $L_{\rm H}$ plane (see Eq. (11)). These empirically determined relations are:

$$% B_{{\rm lim}} - H = -12.7 - 5~{\rm log}({\rm dist}) + 2.5~{\rm log} L_{\rm H}$$ (9)

where $B_{{\rm lim}}=-19.2$ corresponds to the limiting magnitude ( $m_p \leq ~15.7$) at the Coma distance that we assume 96 Mpc.

$$% {\rm log} b = 0.56 - 0.52 (B-H)$$ (10)

$$% {\rm log} b = 7.16 + 2.6~{\rm log}({\rm dist}) - 1.3~{\rm log} L_{\rm H}.$$ (11)

Equation (11) is represented in Fig. 8 with a dashed line. In conclusion, faint-low star forming galaxies at the distance of Coma below the diagonal line of Fig. 8 are severely undersampled.

  .

 

Table 2: The $M_{{\rm gas}}$ vs. diameter relation for isolated galaxies.

Type	a	b	R2
Sa-Sb	7.62	1.55	0.75
Sbc-Sc	7.48	1.68	0.75
Scd-Irr	7.74	1.49	0.77

 

 

Table 3: The newly observed galaxies.

VCC/CGCG	NGC/IC	UGC	${\rm RA}~~(J2000)$	Dec	$m_{{\rm pg}}$	Vel	$T_{{\rm on}}$	$R_{{\rm ON}}$
(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
343	3148	-	121921.68	075213.8	15.1	2479	20	0.79
841	-	-	122547.40	145711.4	15.6	501	20	0.66
15031	4771	8020	125321.85	011613.5	13.3	1119	15	0.85
15049	4845	8078	125801.33	013430.3	12.9	1097	15	0.85
15055	4904	8121	130058.89	-000142.4	13.2	1174	15	0.85
41041	4116	7111	120736.33	024133.1	13.0	1309	15	0.85
43028	4688	7961	124746.67	042005.3	14.5	984	15	0.84
43034	4701	7975	124911.87	032324.5	13.1	727	15	0.79
43054	4765	8018	125314.52	042749.4	13.0	725	15	0.79
69036	4067	7048	120411.46	105114.8	13.2	2424	15	0.8
100015	4758	8014	125244.16	155050.9	14.1	1240	15	0.85
157075	-	-	115940.06	263248.7	15.7	6694	15	0.77
160121	-	8161	130329.11	263300.8	15.5	6676	20	0.77

5.3 The SFR in the Virgo cluster

Figure 9: The relation between the birthrate parameter and the NIR luminosity (mass) for the Virgo galaxies. Empty symbols represent galaxies with normal gas content ( $Def_{{\rm gas}}<0.4$) while deficient galaxies ( $Def_{{\rm gas}}>0.4$) are given with filled symbols.

Figure 10: The relation between the birthrate parameter and the NIR luminosity (mass) for Virgo+Coma galaxies with normal gas content ( $Def_{{\rm gas}}<0.4$).

The selection effect mentioned above affects the Virgo sample to a much lesser extent, because Virgo is 3.7 mag closer than Coma. When we consider the Virgo galaxies alone in Fig. 9 we include dwarf systems with $L_{\rm H}$ fainter by almost 2 orders of magnitudes with respect to Coma. The scatter of the b vs. $L_{\rm H}$ relation increases considerably because the large majority of faint Virgo objects have b lower than Coma. This is in agreement with Kennicutt (1983) who found evidence for significant $\rm H_\alpha$ deficiency in 12 Virgo galaxies with respect to isolated galaxies. Galaxies with b as low as the ones in Virgo might exist in the Coma+A1367 clusters as well, but are not observed because of the previously mentioned observational bias. Thus we conclude that, at any given mass, spirals belonging to the Virgo cluster have their present star formation activity significantly lower than isolated galaxies.

It remains to be explained why. The first thing to explore is whether their gaseous content is sufficient for fueling the star formation. Cluster spirals are in fact known to suffer from HI deficiency (Giovanelli & Haynes 1985; Solanes et al. 2001), a pattern that is interpreted in the framework of the ram pressure mechanism (Gunn & Gott 1972).

When galaxies are separated according to their gas deficiency parameter (see Fig. 9), we recognize that, at any given $L_{\rm H}$, galaxies with "normal'' gas content ( $Def_{{\rm gas}}<0.4$) (open symbols) have their b parameter significantly higher than gas "deficient'' objects.

Figure 11: Histograms of the residual $b_{{\rm obs}}-b_{{\rm mod}}$ for the Coma galaxies (dashed line), for Virgo (continuous line) and for the Virgo galaxies with normal HI content ( $Def_{{\rm gas}}<0.4$) (dashed histogram).

Figure 12: The relation between $b_{{\rm obs}}-b_{{\rm mod}}$ and the $Def_{{\rm gas}}$ parameter. Empty circles are Coma supercluster galaxies, empty squares are "normal'' Virgo clouds (N, W, S), filled circles are "deficient'' Virgo clouds (A, B, E).

Figure 10 is restricted to the non deficient galaxies of both the Virgo and Coma regions. In this and in the previous figures the dotted curve represents $b_{{\rm mod}}$ i.e. the b vs. $L_{\rm H}$ relation expected from the closed-box scenario, in the assumption that $\tau$ is inversely proportional to $L_{\rm H}$ according to Eq. (6). Galaxies in Fig. 10 are found in relatively good agreement with $b_{{\rm mod}}$, in other words their residuals $b_{{\rm obs}}-b_{{\rm mod}}$ are small. This is evidenced in the histograms of Fig. 11 where the distribution of the residuals $b_{{\rm resid}}= b_{{\rm obs}} - b_{{\rm mod}}$ is given separately for the Coma galaxies, for the Virgo galaxies and for the subsample of the Virgo galaxies with normal gas content ( $Def_{{\rm gas}}<0.4$). Large negative residuals, implying a factor of 3 lower SFR, are associated with significantly gas deficient galaxies. It is concluded that, at any given luminosity, the principal parameter regulating the current star formation activity in cluster spirals is the availability of gas at their interior. This is further evidenced in Fig. 12, where $b_{{\rm resid}}$ is plotted against the gas deficiency parameter, showing a significant linear anti-correlation: $b_{{\rm resid}}=0.04$- $0.68 \times Def_{{\rm gas}}$.

Figure 13: The distribution of the "quenched" (empty symbols) and "healthy" (filled symbols) galaxies in the Virgo cluster. Positions of M87 and M49 are shown by crosses.