1 Introduction

For some 150 Myr, the Magellanic-type irregular galaxy NGC1569 experienced a starburst of relatively low average star formation rate ($\sim$0.5$M_{\odot}$ yr^-1), until approximately 10 Myr ago when the starburst gradually ceased. Throughout the body of NGC1569 the starburst has produced many young stars (Greggio et al. 1998; Aloisi et al. 2001) and a large number of star clusters (Hunter et al. 2000), together containing a total mass of $\sim$10⁸$M_{\odot}$. However, evidence for recent and locally very efficient star formation in NGC1569 comes from the Super Star Clusters (SSCs) A and B (Ables 1971; Arp & Sandage 1985; O'Connell et al. 1994) which are similar in size and mass to those found in NGC1705 (Melnick et al. 1985), M 82 (O'Connell et al. 1995; de Grijs et al. 2001), and other amorphous, irregular, and interacting galaxies. The SSCs are sites where in a relatively small volume of 50kpc³ a large number of stars ($\sim$10⁶$M_{\odot}$) have spontaneously formed. Many massive stars produced in the starburst, either in the body of NGC1569 or in and near the SSCs and intermediate-size clusters, already ended in supernova (SN) explosions which created bubbles and kpc-sized loops, an outflow of hot X-ray emitting gas, and a component of young synchrotron radiation (Waller 1991; Heckman et al. 1995; Israel & de Bruyn 1988). We may therefore expect that NGC1569 contains a few radio supernovae (RSNe) and supernova remnants (SNRs), in particular in the region of the SSCs and the intermediate-size clusters where higher than average star formation occurred only $\sim$5-10Myr ago (Prada et al. 1994; O'Connell et al. 1994; Origlia et al. 2001) and where some star formation may still go on, although today the galaxy contains only a few 10⁶$M_{\odot}$ of locally concentrated molecular gas (Greve et al. 1996; Taylor et al. 1999). We have observed NGC1569 with MERLIN at 1.4 and 5 GHz in order to search for RSNe and SNRs.

1.1 NGC 1569 compared to the prototype starburst galaxy M 82

For an investigation of the region around A and B, a comparison with the SSC and SNR population and environment in the prototype starburst galaxy M 82 is highly relevant, in particular regarding the issue whether the RSNe and SNRs in M 82 are associated with SSCs. The starburst in M 82 has produced a large number of SSCs of which $\sim$200 are seen with the HST, both in the active starburst regions "M 82A'' and "C'' (O'Connell et al. 1995; nomenclature from O'Connell & Mangano 1978), and in the more ancient starburst region "B'' just outside the centre (de Grijs et al. 2001). These SSCs are $\ge$5-50 Myr old so that many massive stars have already ended in a SN explosion. Although we may expect that some of the approximately 40, resp. 50 RSNe and SNRs in M 82 detected with MERLIN (Muxlow et al. 1994) and the VLA (Huang et al. 1994) are associated with SSCs, none, or at most one, coincides with the SSCs seen with the HST (Golla et al. 1996). Similarly, none - or at most one - of the $\sim$10 H$_{\alpha }$-bright SNR candidates detected by de Grijs et al. (2000) in M 82B coincides with either bright VLA 8.4 GHz sources, or the optically bright, slightly evolved SSCs found in large numbers in this region (cf. de Grijs et al. 2001). A simple calculation shows that in a population of 100 star clusters of ages similar to those estimated for M 82A and containing 10⁵ and 10⁶ stars, one would expect to detect between about 5 and $\sim$50 type II SNRs at any given moment, assuming any reasonable range of initial mass functions. The question remains, therefore, why none of the optically-detected young compact star clusters show any evidence for the presence of SNRs.

The hypothesis brought forward by Golla et al. (1996) for the absence of RSNe and SNRs in and near SSCs suggests that the visible SSCs of M 82 are located in the foreground and outside appreciable concentrations of interstellar gas so that the SN explosions were unable to sweep up gas and form SNRs. They argue furthermore that there are 1500-3000 SSCs in M 82 and that the detected RSNe and SNRs are hidden behind dense layers of dust so that the associated SSCs are not seen. This argument has apparently gained support from the recent MERLIN observation (Wills et al. 1998) of H I absorption in the direction of many RSNe and SNRs in M 82, and from estimates by Mattila & Meikle (2001) that the MERLIN-detected sources in M 82 are hidden behind dust of $<\!\!A_{V}\!\!>\, = 24$ ($\sigma$$\approx$9) mag extinction. Evidently, under this condition none of the associated SSCs would be visible.

Taking M 82 as example, on statistical arguments we may expect not to find in NGC1569 a short-lived RSN or a SNR in or near the SSCs. If indeed the $\sim$ $3\times10^{8}$$M_{\odot}$ produced in the starburst of M 82 (McLeod et al. 1993) is primarily concentrated in the predicted 1500-3000 SSCs, and if the 40-50 RSNe and SNRs observed today originated in or near SSCs, then at present at most every 1/50th to 1/100th SSC would be associated with a RSN and SNR. Adopting similar conditions for the environment of the SSCs in NGC1569, the chance to observe a RSN or SNR in or near A and B, and in and near the intermediate-size clusters, is extremely small. This comparison however does not consider the possibility that SNRs in a dense gas environment, such as in M 82, may develop differently than in a Magellanic-type galaxy with generally a small amount of gas, such as in NGC1569. Evidence and arguments for different conditions in the interstellar medium in M 82 have for instance been advocated by Pedlar et al. (1999) and Chevalier & Fransson (2001). Finally, we may also argue that the environment in and near SSCs and the intermediate-size clusters may be particularly hostile at least for the formation of SNRs. The matter ejected in a SN explosion in or near the clusters is quickly dispersed because of stellar winds, nearby SN explosions, and the strong gravitational field of the clusters.

Figure 1: 1.4GHz MERLIN observation (contours in steps of 30$\mu$Jy/beam; first negative and positive contour at 50$\mu$Jy/beam) of the field around the super star clusters A and B and the star cluster C (= No. 10) of NGC1569 (see Taylor et al. 1999 for a detailed HST, H$_{\alpha }$ and CO image). The encircled areas around the clusters have a diameter of $\sim$50pc. The crosses show the positions of star clusters identified by Hunter et al. (2000); the open circles, diamonds, and triangles show the positions of WR objects (S,C, and U) observed by Buckalew et al. (2000). 1'' is equivalent to 11pc at the distance of 2.2Mpc. The MERLIN-detected sources M-1,2,3,4,5 (Table 2) and the tentative sources M-a,b,c,d (Table 3) are indicated. The stars show the positions of the VLA-detected sources No.8 and No.16 (vdHGI).

Because of the smaller distance to NGC 1569 (2.2Mpc; Israel 1988) than to M 82 (3.6Mpc; cf. Freedman et al. 1994; Sakai & Madore 1999), 1.4GHz and 5GHz MERLIN observations with a resolution of 200mas ($\sim$2pc) and 50mas ($\sim$0.5pc), respectively, are suitable for a search of RSNe and SNRs. The RSNe and SNRs detected in M 82 with MERLIN (Muxlow et al. 1994; Wills et al. 1997) and the VLA (Huang et al. 1994) are either unresolved or have diameters of up to $\sim$5pc ($\sim$400mas); the flux densities measured at 1.4GHz and 5GHz are between $\sim$0.5mJy and $\sim$20mJy. Similar sizes and flux densities are expected for the RSNe and SNRs in NGC1569, if present at all.