2.1 Host galaxy pixel statistics

SN environment studies started in earnest with the introduction of host galaxy ‘pixel statistics’. In 2006, two papers independently developed very similar techniques with the aim of analysing where within the distribution of pixel counts of particular galaxies the SN environment falls. James & Anderson (Reference James and Anderson2006) investigated the association of different SN types with host galaxy ongoing SF, through narrow-band Hα observations, while Fruchter et al. (Reference Fruchter2006) analysed the association of LGRBs, to host galaxy g′-band light as compared to CC SNe. The majority of the below discussion will focus on the analysis using the James & Anderson method, that will now be described in more detail. However, the Fruchter et al. method is very similar (and has also been used subsequently by e.g. Kelly, Kirshner, & Pahre Reference Kelly, Kirshner and Pahre2008).

The description to provide pixel statistics (dubbed ‘NCR’ values in Anderson & James Reference Anderson and James2008) is outlined below with respect to Hα imaging. However, this description is valid for analysis of host galaxy observations in any waveband (and indeed could be applied to spectral observations as we will outline in Section 5). To begin, continuum-subtracted Hα images are trimmed to remove sky regions outside the galaxy. Then, pixel counts within the resulting image are ordered in terms of increasing count. From this, the cumulative distribution is formed. This cumulative distribution is then normalised by dividing by the total count of all pixels. This then leaves negative values, followed by NCR values between 0 and 1. Negative NCR values, the result of noise in the sky, are then set to 0. Thus, this statistic is made such that values of 0 are equal to sky or zero flux, while a value of 1 means that the pixel contains the highest flux of the whole galaxy. The final NCR value can thus be understood as the fraction of Hα flux arising from pixels with lower counts than the SN-containing pixel. One can then proceed to measure NCR values for different SNe within different galaxies, and build statistical samples to be compared. If a distribution accurately follows the (in this example case) Hα emission, then one expects the distribution to be flat with a mean NCR of 0.5. This would imply that in a population that accurately traces host galaxy Hα emission, an equal number of SNe have NCR values in each bin of NCR (for more details and discussion see Anderson et al. Reference Anderson, Habergham, James and Hamuy2012).

Using pixel statistics, one can use differences in the association of SN types to tracers of different stellar populations to constrain progenitor properties. In the above example of Hα emission, this then gives differences in associations to very young stellar populations, and hence can be used to probe relative age differences between progenitors, as will be shown below.

The spatial resolutions probed by pixel statistics depend on the distance of the host galaxies observed, together with the pixel scale of the instrument used and the image quality of those observations. In the case of Hα NCR statistics, which will dominate the discussion below, the median physical sizes probed are on the order of 300 pc (see Anderson et al. Reference Anderson, Habergham, James and Hamuy2012, Reference Anderson, James, Förster, González-Gaitán, Habergham, Hamuy and Lyman2015 for more detailed estimations). Hence, in general these statistics do not probe individual H ii regions, but larger star-forming regions within galaxies.

2.2 Radial distributions

A more generalised environmental parameter than that of the above pixel statistics, is the radial position within hosts where SNe are found. Within galaxies the properties of stellar populations can change significantly with radial position. Metallicity gradients can exist (see e.g. Henry & Worthey Reference Henry and Worthey1999), in that the central parts of galaxies are generally more metal rich than outer regions. There also exist distinct populations in stellar bulges and bars within the centres of galaxies. Galaxy interactions and mergers can also significantly affect the SF processes within galaxies. Hence, one can analyse differences in the radial distributions of SNe and try to understand these in the context of the different stellar populations found at different galactocentric radii.

The first studies to investigate the radial distributions of SNe generally normalised galactocentric distances to e.g. R ₂₅ (radius of galaxy to the 25th mag. isophote) or worked with absolute distances. One issue with this type of normalisation is that it does not take into account the actual radial distributions of different stellar populations. James & Anderson (Reference James and Anderson2006) analysed the radial distributions of SNe with respect to those of the light as seen through distinct filters. This analysis measures the ellipse with the same parameters as that of their host galaxy that just includes each SN. One then measures the flux contained within this ellipse, and normalises to the flux contained within an ellipse out to distances where the galaxy flux is consistent with sky values. One again obtains values between 0 and 1, where a value of 0 would mean that a particular SN occurs at the central peak of emission, while a value of 1 means that the SN occurs out at distances where no significant galaxy flux is measured. Building radial distributions of different SN types one can then analyse differences and try to tie these to differences in progenitor properties. In the below discussion, we concentrate on SN radial distributions with respect to continuum R-band light, which is referred to as Fr _R. For a full discussion of these analysis methods, see James & Anderson (Reference James and Anderson2006) and Anderson & James (Reference Anderson and James2009). There are several selection effects which may be at play in such an analysis, mostly due to missing SNe in the central parts of host galaxies due to the high surface brightness background. However, as has been shown in Habergham, James, & Anderson (Reference Habergham, James and Anderson2012), there is no significant evidence that these issues will affect one SN type preferentially over the other, especially in the case of differences between SNe II and SNe Ibc. In Section 4.5, we discuss selection effects in general with respect to environment analyses.

2.3 H ii region metallicities

CC SN progenitor metallicity is predicted to play a significant role in the determination of SN type, especially in the single star progenitor scenario (see predictions from e.g. Heger et al. Reference Heger, Fryer, Woosley, Langer and Hartmann2003). In extra-galactic studies, the simplest way to measure stellar population metallicity is through observing the ratio of strong emission lines produced through the ionisation of ISM material by massive stars. The metallicity of choice is that of oxygen, as this is the most abundant metal found in H ii regions and shows strong emission lines in H ii region spectra. In addition, ISM oxygen abundances show good correlation with stellar metallicities of massive stars (see e.g. Simón-Díaz & Stasińska Reference Simón-Díaz and Stasińska2011). While there now exist a plethora of oxygen abundance diagnostics, there is significant disagreement between different diagnostics, meaning that constraints on H ii region metallicities in absolute terms are somewhat problematic. A detailed discussion of these issues, together with suggestions of which diagnostic to use for specific cases is given in Kewley & Ellison (Reference Kewley and Ellison2008). While oxygen abundance measurements between diagnostics can often show large offsets, relative metallicities within a given diagnostic are thought to be much more reliable. Hence, one can investigate H ii region metallicities between different SN types, using a specific diagnostic.

While the ISM metallicity presents the gas abundance at the epoch of observation, given the relatively short lifetimes of CC SNe (at most several 10s of Myrs) one can assume that the SN host H ii region metallicity is a reasonable tracer of progenitor metallicity. One issue is that it is common that at exact explosion sites emission lines are relatively weak, meaning one has to observe the nearest emission line region. Indeed, this is shown in Section 3.1, where particularly for SNe II (but also for many SNe Ib, and a small number of SNe Ic), CC SNe are found to explode within environments with relatively low Hα flux. While it is not clear that this should bring any systematic bias to CC SN metallicity measurements, it is important to note the complications in direct progenitor metallicity constraints from environments. Some studies have attempted to negate this problem by only analysing metallicities which have been extracted at exact explosion sites. However, if there is some progenitor effect which determines whether an SN is less or more likely to explode within a bright H ii region (which we argue there is), then one may be simply replacing one bias with another. Hence, in our analysis below we include all ‘environment’ metallicity measurements. We note that a relatively recent review of metallicity measurements of CC SNe and those SNe accompanying LGRBs, was also presented in Modjaz (Reference Modjaz2011).

There have now been multiple studies of CC SN host H ii region metallicities (Modjaz et al. Reference Modjaz2008; Anderson et al. Reference Anderson, Covarrubias, James, Hamuy and Habergham2010; Modjaz et al. Reference Modjaz, Kewley, Bloom, Filippenko, Perley and Silverman2011; Leloudas et al. Reference Leloudas2011; Modjaz Reference Modjaz2011; Sanders et al. Reference Sanders2012; Kuncarayakti et al. Reference Kuncarayakti2013a, Reference Kuncarayakti2013b; Stoll et al. Reference Stoll, Prieto, Stanek and Pogge2013; Taddia et al. Reference Taddia2013a). However, overall conclusions on progenitor metallicities implied from these results are still not definitive (more details from these studies will be discussed below). In this review, we present a meta-analysis, combining all of the above measurements to try to further probe whether environment metallicity differences between SN types indeed exist. The most common metallicity calibration used is that from Pettini & Pagel (Reference Pettini and Pagel2004), and hence we take those values for our analysis. Apart from being the commonly used diagnostic (and hence published abundance values being readily available), the PP04 metallicity scale also has the advantage that it is derived using emission lines very close in wavelength: Hα together with [N ii], and Hβ together with [O iii]. This means that uncertainties in extinction corrections and/or relative spectral flux calibration are unimportant. The PP04 scale has two metallicity indicators: N2, using only Hα and [N ii], and O3N2, which additionally uses Hβ and [O ii] line strengths. In a significant number of cases publications only list N2 or O3N2 values (the former because in many cases one only detects Hα and [N ii]). As these two diagnostics are on the same scale (which is tied to the ISM electron temperature), we use a combination of both values, preferring O3N2, but using N2 when the former is not available.

In the rest of this review, we will concentrate on analysis and subsequent discussion of the results obtained through the above methods. However, there are specific cases which are not covered by the above and these will be introduced and discussed separately.

3.1 Pixel statistics

In Figure 1, the Hα NCR pixel statistics cumulative distributions of: SNe Ia, SNe II, SNe Ib, and SNe Ic are presented (taken from Anderson et al. Reference Anderson, Habergham, James and Hamuy2012). These plots (which will be repeated in various forms below) can be understood in the following way. As outlined above, if a population accurately traces the emission in which an analysis is presented—in the case of Figure 1 Hα emission—then one expects that distribution to be flat with a mean of 0.5. On cumulative distribution plots, this translates to being consistent with a straight diagonal line which crosses the plot with intersections at 0,0 and 1,1. As one moves away to the upper left of this diagonal (e.g. see the SNe Ia in Figure 1), then a population is displaying a lower degree of association to the emission. If a population falls below this diagonal line to the right, then this equates to a distribution that is biased towards the peaks of the emission, and does not simply follow the emission on a one-to-one basis. Following this we see that the SNe Ia show the lowest degree of association to the line emission, followed by SNe II, SNe Ib, and finally SNe Ic showing the largest degree of association. Hα emission arises from recombination of hydrogen ionised by the flux of nearby massive stars. Hence, where one observes a concentration of Hα emission, an H ii region, one is in effect observing a concentration of young massive stars, where the dominant contributors to the ionising flux are thought to be stars with initial masses ⩾ 15–20 M_⊙. Hence, these regions within galaxies are thought to have ages < 15 Myrs (see e.g. Kennicutt Reference Kennicutt1998). The distribution of Hα flux within galaxies therefore traces ongoing SF. Hence, if different SNe show differences in their association to the line emission, as is shown in Figure 1, then the most straightforward interpretation of this is that we are observing differences in the mean progenitor ages, and therefore masses of SN classes. Through this interpretation, we can then order SNe in terms of progenitor age and mass. The results of Figure 1 imply that, as expected, SNe Ia have the longest progenitor lifetimes and hence the lowest masses. SNe II are next, and appear to have the lowest mass progenitors of the CC population, consistent with constraints from direct progenitor detections. Then follows the SNe Ib, where pixel statistics suggest a slightly higher association to ongoing SF than SNe II, however we note that this difference is marginal. The SNe Ic meanwhile show an almost perfect one-to-one association to Hα emission. Hence, in this interpretation we have an increasing progenitor mass sequence of: SNe Ia → SNe II → SNe Ib → SNe Ic. We note that there has been considerable criticism of this simple interpretation in the literature (see e.g. Crowther Reference Crowther2013; Smith & Tombleson Reference Smith and Tombleson2015), and we return to these issues later in Section 4. However, it is important to emphasise (irrespective of interpretations) the key results from this analysis.

Figure 1. Cumulative Hα NCR distributions of the main SN types: SNe Ia, SNe II, SNe Ib, and SNe Ic (the combined SN Ibc population is also shown). The straight black diagonal line represents a hypothetical distribution, infinite in size, which accurately traces (in this case) host galaxy on-going SF as traced by Hα line emission. This plot is a reproduction of Figure 2 from Anderson et al. (Reference Anderson, Habergham, James and Hamuy2012), but here we remove the SNe IIb and SNe IIn from the SN II sample, to be consistent in our sample analysis throughout the review.

1. 1. Neither SNe II nor SNe Ib trace ongoing SF as traced by Hα emission.

2. 2. SNe Ic accurately trace the spatial distribution of ongoing SF within hosts.

3. 3. SNe Ib and SNe Ic show a statistically significant difference in their association with ongoing SF.

At first glance, it may seem somewhat surprising that the SN II population does not follow the spatial distribution of Hα emission, given that the latter traces high-mass SF, and that SNe II are the explosions of massive stars. However, as outlined above, Hα emission traces SF on timescales of < 15 Myrs, while the progenitor population of SNe IIP has been estimated to have masses of ~ 8–16 M_⊙ (Smartt et al. Reference Smartt, Eldridge, Crockett and Maund2009). Hence, the latter have delay times before explosion longer than the lifetimes of typical H ii regions. Near-UV emission traces SF on longer timescales of up to ~ 100 Myrs (e.g. Gogarten et al. Reference Gogarten2009b). It has been shown (Anderson et al. Reference Anderson, Habergham, James and Hamuy2012), that indeed the SN IIP population accurately traces the near-UV emission within their host galaxies in terms of NCR pixel statistics, as is shown in Figure 2. This is completely consistent with these SNe arising from RSG progenitor stars at the low end of the CC SN mass range, as constrained by direct progenitor detections.

Figure 2. Cumulative Hα and near-UV NCR distributions of SNe IIP. Reproduction of Figure 4 from Anderson et al. (Reference Anderson, Habergham, James and Hamuy2012).

Similar work on pixel statistics has been published by other authors. Kelly et al. (Reference Kelly, Kirshner and Pahre2008) presented g′-band pixel statistics for SNe and compared these to LGRBs (these authors used the fractional flux formalism of Fruchter et al. Reference Fruchter2006, rather than the NCR method outlined above). These distributions are shown in Figure 3. While Hα line emission within galaxies is dominated by contributions from ongoing massive SF, the g′-band light is a much coarser tracer of young populations. However, it is interesting that similar trends to the above are seen in terms of the relative differences between populations. SNe II follow the g′-band light on an almost one-to-one basis, while the SNe Ib appear to be slightly more clustered on the brighter pixels. SNe Ic meanwhile are much more concentrated on the brightest g′-band pixels (implied by the fact that the distribution falls to the bottom right of the diagonal). In absolute terms, this clustering of SNe Ic (and to a lesser extent SNe Ib) on the brighter g′-band regions perhaps implies that massive SF is clustered with the majority of it being found at the brightest continuum (g′-band) peaks. One of the main conclusions of Kelly et al. was that SNe Ic and LGRBs have similar mass progenitor stars due to their similar associations to host galaxy young stellar populations as traced by g′-band light. These same statistics were used by Raskin et al. (Reference Raskin, Scannapieco, Rhoads and Della Valle2008) and compared to galaxy models to constrain CC SN progenitor ages. It was concluded that the association of SNe Ic to g′-band light implied a minimum progenitor mass for SNe Ic of 25 M_⊙.

Figure 3. Cumulative g′-band fractional flux distributions of SNe and long-duration Gamma Ray Bursts. This figure is Figure 2 from Kelly et al. (Reference Kelly, Kirshner and Pahre2008), and is reproduced with permission of the AAS. We thank Pat Kelly for consent to use this figure.

Kangas et al. (Reference Kangas, Mattila, Kankare, Kotilainen, Väisänen, Greimel and Takalo2013) investigated specifically the local environments of CC SNe within infrared-bright galaxies. Using NCR pixel statistics, they presented similar distributions and relative differences in distributions between SNe II and SNe Ic populations to the Anderson et al. results. Interestingly, these authors also presented pixel statistics with respect to other wave-bands. In particular, they showed that while SNe II did not follow the K-band light of their host galaxies, the SNe Ic do. This is surprising given that historically K-band light has been thought to trace the old stellar mass within galaxies. Kangas et al. suggested that this was due to the fact that in starburst environments the K-band light is dominated by RSGs, and hence at least in their sample is a tracer of SF. It is also interesting to note that with respect to the near-UV emission SNe Ic cluster onto the brightest pixels and show more than a one-to-one relation to the recent SF. This is also seen in Figure 19 presented below in Section 3.6.1, and may imply that the most intense SF happens in the brightest UV regions of galaxies.

Finally, we note that Crowther (Reference Crowther2013) and Galbany et al. (Reference Galbany2014) have also investigated the association of SNe with SF regions, through pixel statistics and by measuring distances to the nearest H ii region. Both authors find similar statistics to those that have been presented above: SNe Ibc show a higher association to ongoing SF than SNe II.

3.2 SN radial distributions

An updated sample of radial distributions of CC SNe with respect to host galaxy R-band light was published in Habergham et al. (Reference Habergham, James and Anderson2012, following Anderson & James Reference Anderson and James2009, and Habergham, Anderson, & James Reference Anderson, Covarrubias, James, Hamuy and Habergham2010), also see comparisons to the radial distribution of SNe Ia in Anderson et al. Reference Anderson, James, Förster, González-Gaitán, Habergham, Hamuy and Lyman2015). That analysis concentrated on the radial distributions of the main CC SN types of SNe II, SNe Ib, and SNe Ic. It has long been observed that SNe Ibc are more centrally concentrated within hosts than SNe II (see e.g. van den Bergh Reference van den Bergh1997 for an early example). Historically, this population difference was interpreted as a progenitor metallicity effect (which was also the initial conclusion in Anderson & James Reference Anderson and James2009), with a higher rate of SNe Ibc with respect to SNe II in the higher metallicity central parts of galaxies. In the single star progenitor scenario, this is predicted to be the case, as more metal rich stars have stronger radiatively driven winds (e.g. Puls et al. Reference Puls1996; Kudritzki & Puls Reference Kudritzki and Puls2000; Mokiem et al. Reference Mokiem2007), and therefore lose a greater proportion of their envelopes and hence explode as SNe Ibc rather than SNe II. This conclusion has also been further supported by global host galaxy studies, where SNe Ibc are found to occur in more massive galaxies than SNe II (Prantzos & Boissier Reference Prantzos and Boissier2003; Boissier & Prantzos Reference Boissier and Prantzos2009, although see Arcavi et al. Reference Arcavi2010 for how this is complicated when one separates SNe into their further sub-types), presumed to be more metal rich, and higher metallicity galaxies through direct global measurements (Prieto, Stanek, & Beacom Reference Prieto, Stanek and Beacom2008a). However, when one separates host galaxies by their degree of disturbance, metallicity as the dominant factor in explaining radial distributions of CC SNe becomes untenable.

Separating CC SN host galaxies by their degree of disturbance was first achieved by Habergham et al. (Reference Habergham, Anderson and James2010), and a further analysis and discussion was presented in Habergham et al. (Reference Habergham, James and Anderson2012). This classification separated host galaxies into three groups: undisturbed hosts, disturbed hosts, and an extreme sample which is generally dominated by major mergers (see examples of this last group in Figure 4). Galaxy classification was achieved using a series of selection criteria such as asymmetries, the appearance of a close companion, etc. The reader is referred to Habergham et al. (Reference Habergham, James and Anderson2012) for a full discussion. Here, we add a further 67 CC SNe to the sample analysed in Habergham et al. (Reference Habergham, James and Anderson2012), including SNe from the Carnegie Supernova Project (CSP), where we analyse host galaxy r′-band images originally obtained for light-curve host galaxy subtraction, and those SNe analysed in Kangas et al. (Reference Kangas, Mattila, Kankare, Kotilainen, Väisänen, Greimel and Takalo2013) who employ the same method as Anderson et al. (Reference Anderson, Habergham, James and Hamuy2012). The final sample discussed here is of 347 CC SNe. This is split into: 226 SNe II (where we include those SNe classified as ‘II’, ‘IIP’, and ‘IIL’), 50 SNe Ib, 65 SNe Ic, and a total of 121 SNe Ibc. We exclude from this analysis both SNe IIb and SNe IIn (see Habergham et al. Reference Habergham, Anderson, James and Lyman2014 for a discussion of the radial distribution of SNe IIn).

Figure 4. Examples of the most disturbed galaxy systems where CC SNe have exploded in the very central regions. This figure is Figure 5 from Habergham et al. (Reference Habergham, Anderson and James2010), and is reproduced with permission of the AAS.

For each SN in the above sample, we calculate an Fr _R which gives a measure of where within the radial distribution of R-band light of their host galaxies they exploded. The sample is then split by the degree of disturbance, and we plot the resulting histograms and cumulative distributions in Figures 5 through 7. In Table 1, we give the mean Fr _R values for each population in each of the three galaxy samples, together with their overall values. Also presented are the KS-test statistics which give ‘D’ parameter indicating the amplitude of the difference between two populations and the probability ‘p’ that the SNe II and SNe Ibc distributions are drawn from the same parent population.

Figure 5. Histograms of the Fr _R distributions of CC SNe in undisturbed galaxy systems. In the bottom panel, the SN II population is presented, and in the middle panel the SNe Ibc. In the top panel, the cumulative distributions of both populations are presented.

Figure 6. Histograms of the Fr _R distributions of CC SNe in disturbed galaxy systems. In the bottom panel, the SN II population is presented, and in the middle panel the SNe Ibc. In the top panel, the cumulative distributions of both populations are presented.

Figure 7. Histograms of the Fr _R distributions of CC SNe in the most disturbed galaxy systems. In the bottom panel, the SN II population is presented, and in the middle panel the SNe Ibc. In the top panel, the cumulative distributions of both populations are presented.

Table 1. CC SNe Fr _R distributions and KS-test statistics.

In column 1, the SN distribution type is listed, followed by the number of events analysed in column 2. Then, the mean Fr _R value is listed for each distribution. For each set of two distributions, the KS-test ‘D’ and probability value ‘p’ are listed in columns 4 and 5 respectively.

Analysing Figures 5 through 7 and Table 1, we see the following trends. In all cases, the SN Ibc population is more centralised within host continuum R-band light than the SNe II. However, this centralisation becomes more apparent as one moves from the undisturbed through the disturbed and finally in the extreme disturbed samples. This can be seen in the plots but also in the KS-test ‘D’ values in Table 1 which increase as one moves from undisturbed to disturbed host galaxy samples. The figures show that this is mainly driven by an excess of SNe Ibc in the centres of disturbed galaxies. This is to say: as a galaxy sample becomes more disturbed, SNe Ibc become more centralised within their host galaxy light. It is also interesting to note the decrease in the ratio of SNe II to SNe Ibc between the galaxy samples. In the undisturbed sample the ratio is 2.3, then 1.7 in the undisturbed sample, and finally 1.2 in the extreme sample, i.e. the number of SNe Ibc to SNe II increases significantly as the degree of host galaxy sample disturbance increases. This is consistent with previous work (Habergham et al. Reference Habergham, Anderson and James2010, Reference Habergham, James and Anderson2012; Hakobyan et al. Reference Hakobyan2014).

Throughout this review we show that there exist significant differences in the environments of SNe Ib and SNe Ic. In Figure 8, we present the Fr _R distributions for the individual SN Ib and SN Ic populations when the galaxy sample is split into undisturbed and disturbed galaxies (an extreme sample is not discussed here due to the low number of events in that distribution). In Table 2, the mean values and KS-test statistics between these samples are listed. It is observed that here the difference between populations is more pronounced in the undisturbed samples (while above the difference between SNe II and the overall SN Ibc samples was more pronounced in the disturbed and extreme samples). SNe Ic are found to occur more centrally than SN Ibc in both undisturbed and disturbed galaxy samples, however the effect is significantly larger in the former.

Figure 8. Cumulative Fr _R distributions for SNe Ib and SNe Ic after host galaxies have been split into undisturbed and disturbed samples. (Note, here we do not plot the separate extreme sample due to a lack of statistics, but this also forms part of the disturbed sample).

Table 2. SN Ib and SN Ic Fr _R distributions and KS-test statistics.

In column 1, the SN distribution type is listed, followed by the number of events analysed in column 2. Then the mean Fr _R value is listed for each distribution. For each set of two distributions, the KS-test ‘D’ and probability value ‘p’ are listed in columns 4 and 5 respectively.

While SNe Ibc are more centrally concentrated than SNe II in disturbed systems, the explanation of this finding is not simple, and it is not clear that metallicity effects are responsible. While metallicity gradients are observed in most galaxies, they can become much weaker or even non-existent in disturbed/merging galaxies (see e.g. Kewley et al. Reference Kewley, Rupke, Jabran Zahid, Geller and Barton2010 and most recently Sánchez et al. Reference Sánchez2014). During the process of interaction, pristine gas flows to the inner parts of the system, enriching the central parts of galaxies with low abundance material (see e.g. Hibbard & van Gorkom Reference Hibbard and van Gorkom1996; Rampazzo et al. Reference Rampazzo, Plana, Amram, Bagarotto, Boulesteix and Rosado2005; Kewley, Geller, & Barton Reference Kewley, Geller and Barton2006). In other words, metallicity gradients are stronger in undisturbed systems, and hence any progenitor metallicity effect between SNe II and SNe Ibc should be stronger in undisturbed systems than disturbed ones. We also note that Sánchez et al. (Reference Sánchez2014) showed that metallicity gradients can in fact invert in the very central regions, which can further complicate the interpretation of SN radial distributions.

When we separate the SNe Ibc into Ib and Ic classes, we see that the results are reversed. Now, the differences between the radial positions are much larger in undisturbed systems (indeed it is insignificant in disturbed systems). In this case, the most likely cause is indeed environment metallicity, with SNe Ic preferring to explode in the higher metallicity central regions of galaxies with significant metallicity gradients (consistent with the H ii region metallicity results presented below). Before moving on, we reiterate the main points from this radial analysis:

1. 1. Differences between the distributions of Fr _R values of SNe II and SNe Ibc increase as one moves from undisturbed to disturbed host galaxy samples.

2. 2. In disturbed/interacting systems, SNe Ibc are much more centrally concentrated within galaxies than SNe II.

3. 3. There is a central excess of SNe Ibc in the central regions of disturbed galaxies compared to host galaxy continuum light.

4. 4. Differences in the radial distributions of CC SNe cannot be dominated by progenitor metallicity effects.

5. 5. However, in the undisturbed sample SNe Ic are seen to be more centrally concentrated than SNe Ib, hinting at a metallicity effect between these two sub-types.

In Habergham et al. (Reference Habergham, Anderson and James2010) and Habergham et al. (Reference Habergham, James and Anderson2012), it was speculated on how these results can thus be understood after discounting progenitor metallicity effects. Indeed, it was suggested that the centralisation of SNe Ibc in disturbed systems could indicate that in these dense violent SF processes induced in the central regions of these galaxies, that the initial mass function (IMF) is biased to the production of more massive stars, and hence a higher proportion of SNe Ibc to SNe II is produced. It is also possible that in the central regions of these galaxies there is a higher fraction of interacting binaries, and hence a higher number of SN Ibc explosions. Both of these possibilities were also considered in the radial analysis of Hakobyan et al. (Reference Hakobyan, Mamon, Petrosian, Kunth and Turatto2009), and are further discussed below.

Kangas et al. (Reference Kangas, Mattila, Kankare, Kotilainen, Väisänen, Greimel and Takalo2013) also analysed CC SN host galaxies using the same radial analysis as discussed above. These authors concentrated on SNe within infrared bright galaxies (many of which are starbursts), finding very similar results our analysis where we combine their sample into our large radial analysis database. A further interesting result from Kangas et al. is that the SNe II in their sample were found to prefer to explode in the outer regions of the K-band light of their hosts.

3.3 Host H ii region metallicities

The first statistical study of SN environmental metallicities concentrated on probing differences between BL SNe Ic that had or had not accompanied LGRBs, finding lower oxygen abundances for the latter (Modjaz et al. Reference Modjaz2008). This suggests that low metallicity is an important parameter in producing these extreme events. In Anderson et al. (Reference Anderson, Covarrubias, James, Hamuy and Habergham2010), a comparison was made between SNe II and SNe Ibc environment oxygen abundances. It was concluded that any implied progenitor metallicity difference between these SN types was relatively small. That study also suggested that the difference between SNe Ib and SNe Ic H ii region metallicities was small. However, this was contested by Modjaz et al. (Reference Modjaz, Kewley, Bloom, Filippenko, Perley and Silverman2011), who claimed a difference in that SNe Ic arise from regions of higher oxygen abundance. These findings were interpreted as evidence for more complete envelope-stripping in high-metallicity progenitors, indicating a dominant role for line-driven winds. Leloudas et al. (Reference Leloudas2011) also presented an analysis of SNe Ibc environment oxygen abundances finding something in between those two previous studies. A sample of SNe Ibc formed from exclusively non-targeted SN searches was presented by Sanders et al. (Reference Sanders2012). Those authors concluded that only a small difference between SNe Ib and SNe Ic environment oxygen abundance existed, and also analysed previous work in the literature to conclude that any difference between SNe II and SNe Ibc H ii region oxygen abundances was small. A sample of SNe II from the untargeted search program PTF (Palomar Transient Factory, Rau et al. Reference Rau2009) was published by Stoll et al. (Reference Stoll, Prieto, Stanek and Pogge2013). This study looked in detail at the properties of their sample, and found that it had a very similar oxygen abundance distribution to SN II samples drawn from galaxy targeted surveys (i.e. that in Anderson et al. Reference Anderson, Covarrubias, James, Hamuy and Habergham2010). This is somewhat surprising, as it is generally assumed that a sample drawn from targeted searches will be biased towards SNe in more massive, metal-rich host galaxies, compared with those found within untargeted rolling searches. Most recently Kuncarayakti et al. (Reference Kuncarayakti2013a) and Kuncarayakti et al. (Reference Kuncarayakti2013b) published SN II and SN Ibc samples of H ii region oxygen abundances, where they focussed on very nearby events and analysed integral field unit (IFU) spectroscopy, attempting to identify specific parent stellar clusters. Again, they found only marginal evidence for an implied metallicity difference between the two SN samples, although SNe Ic were found to occur in regions of higher oxygen abundance than SNe Ib.

In light of the significant number or studies above, we proceed to present a meta-analysis of all CC SN H ii region oxygen abundance measurements. We also include unpublished values from the CALIFA survey, which will be presented in L. Galbany et al. (in preparation). In our final sample, we combine the unpublished values from CALIFA with those from: Modjaz et al. (Reference Modjaz2008), Thöne et al. (Reference Thöne2009), Anderson et al. (Reference Anderson, Covarrubias, James, Hamuy and Habergham2010), Modjaz et al. (Reference Modjaz, Kewley, Bloom, Filippenko, Perley and Silverman2011), Leloudas et al. (Reference Leloudas2011), Anderson, Habergham, & James (Reference Anderson, Habergham and James2011), Prieto et al. (Reference Prieto2012), Habergham et al. (Reference Habergham, James and Anderson2012), Van Dyk et al. (Reference Van Dyk2012), Sanders et al. (Reference Sanders2012), Stoll et al. (Reference Stoll, Prieto, Stanek and Pogge2013), Tomasella et al. (Reference Tomasella2013), Kuncarayakti et al. (Reference Kuncarayakti2013a, Reference Kuncarayakti2013b), Inserra et al. (Reference Inserra2013), and Taddia et al. (Reference Taddia2013b)Footnote ³. Our final sample is of 245 CC SN host H ii region oxygen abundances. Here, we only choose to analyse SNe II (120 SNe, where this includes SNe IIP, SNe IIL, and those events classified as simply SNe II in the literature), SNe Ib (56) and SN Ic (59, with a total of 125 SNe Ibc), to be consistent with the previous pixel statistics and radial analyses presented above. It is noted that here we include data from many different sources, where data reduction and spectral extraction may be different between samples. These samples also come from different SN search sources, including targeted and un-targeted searches, which again brings issues such as differences in relative ratios of SN types. However, our aim here simply to bring all the relevant data together in one place, and use significant statistics to investigate whether there indeed exist any statistical differences in implied host H ii region metallicities.

In Figure 9, we present the cumulative oxygen abundance distributions of the main CC SN types. All mean values, and KS-test D and p values are listed in Table 3. It is observed that while essentially all distributions have statistically very similar implied environment metallicities, the SN Ic population has the highest mean value. Statistically speaking, we find no evidence for environment and hence progenitor metallicity differences between SNe II and the overall SN Ibc population. We note that there is a suggestion that SNe Ic have higher metallicities than SNe Ib, but this is not statistically significant. In Figure 10, we show the ratio of SNe II to SNe Ibc as a function of host H ii region oxygen abundance. Here, we simply split the sample into five equally-sized bins (in terms of numbers of SNe). Errors on ratios are Poissonian. The ratio is essentially flat with metallicity, with some suggestion that the SN Ibc/SN II ratio decreases in the lower bin (which has a mean 12+log(O/H) of 8.29). In Figure 11, we further split the sample and show how the ratios: SN Ib/II, SN Ib/Ic, and SN Ic/II change as a function of host H ii region oxygen abundance. The SN Ib/Ic ratio appears to increase slightly with decreasing metallicity. The SN Ic/II ratio is flat with oxygen abundance, while the SN Ib/II ratio appears to increase slightly. Again, we emphasise the low significance of all of these trends. We list the important results from this CC SN H ii region metallicity analysis:

1. 1. Overall we find only small, statistically insignificant differences in the environment oxygen abundances of SNe II, SNe Ib, and SNe Ic.

2. 2. There is a suggestion that the ratio of SNe II to SNe Ibc increases in the lowest metallicity bin.

3. 3. The SN Ib to SN Ic ratio appears to increase with decreasing oxygen abundance.

Figure 9. Cumulative distributions of CC SN host H ii region oxygen abundance. Here, we show the SN II, SN Ib, SN Ic, and the combined SN Ibc populations.

Figure 10. The ratio of SN Ibc to SN II events as a function of host H ii region oxygen abundance. The full sample of CC SNe are simply split into five bins of equal size (in terms of number of measurements) of oxygen abundance.

Figure 11. Top: the ratio of SNe Ib to SNe Ic as a function of host H ii region oxygen abundance; Middle: the ratio of SNe Ic to SNe II; Bottom: the ratio of SNe Ib to SNe II. In all three histograms, the samples are simply split into three equal-sized bins of oxygen abundance.

Table 3. CC SN H ii region oxygen abundances and KS-test statistics.

In the top panel in column 1, the SN distribution type is listed, followed by the number of events analysed in column 2. Then the mean Z value is listed for each distribution. In the bottom panel, column 1 indicates the two distributions being compared, and columns 2 and 3 list the KS-test D and p values between those distributions.

These results suggest that progenitor metallicity is not a driving parameter in producing the majority of CC SN diversity. This obviously does not rule out metallicity playing a part in producing specific SN sub-types, or changing the properties of more specific transient features (other than simple type classifications). However, this does seem to suggest that other parameters such as progenitor mass and/or the presence and influence of a binary companion are much more dominant. Indeed, given that single stellar models (see Heger et al. Reference Heger, Fryer, Woosley, Langer and Hartmann2003; Eldridge & Tout Reference Eldridge and Tout2004; Georgy et al. Reference Georgy, Meynet, Walder, Folini and Maeder2009; Ibeling & Heger Reference Ibeling and Heger2013) predict a strong dependence SN type production with respect to progenitor metallicity, the above results (and those published in the literature previously, see Anderson et al. Reference Anderson, Covarrubias, James, Hamuy and Habergham2010; Sanders et al. Reference Sanders2012 for combined SNe II and SNe Ibc analyses) indicate that a significant fraction of CC SNe are the result of massive star binary interactions. We return to this conclusion in Section 4.

3.4 Further environment studies

In this section, we summarise and discuss CC SN environment studies which do not fall easily into previous sections. IFU spectrographs allow one to obtain both spatial and spectral information simultaneously. In Section 5, we discuss the future of environment studies where we argue that IFUs will play a dominant role, following surveys such as CALIFA (Sánchez et al. Reference Sánchez2012), and using new facilities such as the recently commissioned MUSE instrument at the VLT (Bacon et al. Reference Bacon, McLean, Ramsay and Takami2010). Kuncarayakti et al. (Reference Kuncarayakti2013b) and Kuncarayakti et al. (Reference Kuncarayakti2013a) presented IFU environment studies of 11 SNe II and 13 SNe Ibc respectively. They used the power of IFU observations to first identify the nearest stellar cluster to SN explosion positions, before extracting the spectrum at that position. Environment oxygen abundances were then estimated using the same strong emission line diagnostics as outlined above finding similar distributions for SNe II and SNe Ibc. These authors also measured Hα EWs which they used to trace the age of the local stellar population, which can then be translated to progenitor masses using stellar evolution models. In Figure 12, we reproduce the mass–metallicity figure from Kuncarayakti et al. (Reference Kuncarayakti2013b), where measurements were compared to both model predictions and SN II direct progenitor detections. It was concluded that a number of SNe Ibc were associated with stellar clusters with ages consistent with relatively low progenitor masses, where the implied progenitor scenario would be binary systems. In addition, in some cases relatively high progenitor masses were derived for SNe II. In Figure 13, we reproduce the Hα EW distributions from Kuncarayakti et al. (Reference Kuncarayakti2013b), which show differences between SNe IIP, SNe IIL, SNe Ib, and SNe Ic. While the statistics are limited, it appears that Hα EWs are generally similar between SNe IIP and SNe Ibc, with some suggestion that those for SNe Ic are slightly larger, possibly consistent with the Hα pixel statistics presented above. The EWs of SNe IIL appear to have a tendency for very large values, implying high-mass progenitors. However, we caution that these studies are currently limited by low statistics, where issues such as associating a particular cluster to a parent progenitor population may become problematic. A significant number of new observations have been obtained for this study, and results from a larger sample promise to be enlightening. The Kuncarayakti et al. studies concentrate on very nearby SNe where one is resolving individual stellar clusters. Galbany et al. (Reference Galbany2014) used lower resolution IFU observations to also measure Hα EWs at the explosion sites of SNe (and in addition analysed various measures of the association of SNe to SF region, see below). In Figure 14, Hα EW distributions for SNe Ia, SNe II, and SNe Ibc are displayed. As expected, SN Ia environments have the lowest EW distribution. It is interesting that here the SNe Ibc appear to explode in regions of slightly higher EW than SNe II, indicating younger ages, and therefore slightly higher masses for the former. Again, larger samples and further analysis are needed to build on these studies.

Figure 12. Progenitor age against progenitor metallicity plot. SN environment ages and metallicities are derived using IFU spectroscopy. These are compared to a model grid taken from Georgy et al. (Reference Georgy, Meynet, Walder, Folini and Maeder2009), and also SN II direct progenitor detections from Smartt et al. (Reference Smartt, Eldridge, Crockett and Maund2009). This figure is Figure 17 taken from Kuncarayakti et al. (Reference Kuncarayakti2013b) and is reproduced by permission of the AAS.

Figure 13. Hα EWs of the stellar population observed within the parent stellar clusters of CC SNe, derived from IFU spectroscopy. SN IIP, IIL, Ib, and Ic distributions are presented. A larger EW indicates a younger age. This figure is an edited version of Figure 18 taken from Kuncarayakti et al. (Reference Kuncarayakti2013b) and is reproduced by permission of the AAS.

Figure 14. Hα EWs of the stellar population observed at the explosion sites of SNe, derived from IFU spectroscopy. SN Ia, II, and Ibc distributions are presented. This figure is part of Figure 13 taken from Galbany et al. (Reference Galbany2014) and is reproduced by permission of A&A.

Another distinct environment analysis which has yet to be discussed, is multi-colour photometric analysis. Kelly & Kirshner (Reference Kelly and Kirshner2012) presented a large statistical analysis of 519 CC SN environments deriving various parameters for the unresolved stellar populations found at explosion sites. In Figure 15, we present their environment u′-band surface brightness against u′ − z′ colour plot. SNe Ic BL (broad-line) and SNe IIb were observed to explode in significantly bluer regions of their hosts than other SN types, while the u′-band surface brightness of SN Ic and SN Ib environments was observed to be higher than those of SNe II, again somewhat consistent with the Hα pixel statistics above. These differences in environments suggest some differences in progenitor properties such as age or mass (possibly related to u′-band surface brightness) and metallicity (relating to environment colour). Another interesting result from Kelly & Kirshner (Reference Kelly and Kirshner2012) was that no significant difference was found in the environment properties of SNe IIn and the ‘normal’ SN II population. This argues for similar progenitor properties, a conclusion that will be further strengthened below. Kelly & Kirshner presented a number of other interesting environment and global host galaxy measurements, and the reader is encouraged further to explore that publication.

Figure 15. u′-band surface brightness against u′ − z′ colours of CC SN environments. The larger inset shows the two parameters plotted against each other, while above and to the right are the cumulative distributions of each property. This figure is Figure 2 taken from Kelly & Kirshner (Reference Kelly and Kirshner2012), and is reproduced by permission of the AAS. We thank Pat Kelly for consent to use this figure.

3.5 SN Ia environments

This review concentrates on the environmental properties of CC SNe, as those have been the most widely studied. However, especially in the most recent years work has been published on the local environments of SNe Ia. Given the much longer expected delay times of SNe Ia, interpreting environmental results becomes more difficult as progenitors have possibly travelled large distances before explosion. However, some intriguing results have been published.

Raskin et al. (Reference Raskin, Scannapieco, Rhoads and Della Valle2009) used a modified version of the Fruchter et al. (Reference Fruchter2006) fractional flux method to analyse SN Ia environment pixel statistics, comparing observations to analytical galaxy models, concluding that even the ‘prompt’ component of the SN Ia population has delay times of several 100 Myrs. Stanishev et al. (Reference Stanishev, Rodrigues, Mourão and Flores2012) presented a first IFU study of SN host galaxies, analysing the spatial and spectral information obtained for a sample of six SNe Ia, and showing the capabilities of using such observations. Rigault et al. (Reference Rigault2013) also used IFU observations of a large sample of SN Ia host galaxies. These authors concentrated on measuring the strength or presence of Hα emission at the explosion sites of SNe Ia. They found that SNe where emission was indeed detected have redder colours than those where no emission is detected (a similar result was also found by Anderson et al. Reference Anderson, James, Förster, González-Gaitán, Habergham, Hamuy and Lyman2015, see below). Wang et al. (Reference Wang, Wang, Filippenko, Zhang and Zhao2013) speculated that two distinct populations of SN Ia progenitors exist, with those events having higher ejecta velocities being more centrally-concentrated than their lower-velocity counterparts. However, this result has recently been questioned in an independent study by Pan et al. (Reference Pan2015). Rigault et al. (Reference Rigault2013) showed that SNe Ia associated with local Hα emission are more homogeneous and give a lower dispersion when used as distance indicators (also see Rigault et al. Reference Rigault2015). Subsequent work by Kelly et al. (Reference Kelly and Kirshner2015) identified a sub-set of SNe Ia which exploded in environments of high UV surface brightness and SF surface density, which gave significantly more accurate distance measurements than SNe Ia in all environments.

In Anderson et al. (Reference Anderson, James, Förster, González-Gaitán, Habergham, Hamuy and Lyman2015), the measurements described in Sections 2.1 and 2.2 are used to study a sample of ~ 100 literature SNe Ia. Host galaxy NCR pixel statistics were obtained from near-UV through optical and near-IR imaging. It was found that the SN Ia population in star-forming galaxies most accurately follows the spatial distribution of host B-band light, as shown in Figure 16. This implies that the dominant progenitor population of SNe Ia in star-forming galaxies has neither very long delay times, i.e. they do not follow the near-IR J- or K-band light, nor very short delay times, i.e. they do not follow the near-UV emission. In addition, it is shown in Figure 17 that when the analysed SN Ia sample is split by SN colour at maximum, ‘redder’ events are more likely to be associated with Hα emission that their ‘blue’ counterparts. This hints at host galaxy extinction driving the colour diversity of SNe Ia. The reader is encouraged to read that paper in full for an overall view of SN Ia environments.

Figure 16. Cumulative NCR pixel distributions for SNe Ia at distinct wave-band observations: Hα, near-UV, B-, R-, J- and K-band. This figure is taken from Anderson et al. (Reference Anderson, James, Förster, González-Gaitán, Habergham, Hamuy and Lyman2015).

Figure 17. Cumulative Hα NCR distributions for SNe Ia when the sample is split by SN colour at maximum light. The SN II distribution is also shown for reference. This figure is taken from Anderson et al. (Reference Anderson, James, Förster, González-Gaitán, Habergham, Hamuy and Lyman2015).

3.6 The environments of SN sub-types

Above we have focussed discussion on the environments of the main SN types. In such studies, one needs sufficient statistics to enable comparisons between different SN populations to be enlightening. However, some studies have looked at (observed) rarer SN types and studied their environments with the aim of providing progenitor constraints.

3.6.1 Interacting transients

SNe IIn have historically been assumed to arise from extremely massive stars, given the large amounts of CSM material implied from their interaction-driven spectra. Indeed, it has been speculated (e.g. Smith Reference Smith, Kwok and Sanford2008) that their progenitors are luminous blue variable stars (LBVs). In addition, in the case of SN 2005gl the progenitor star was constrained to be a very massive and probable LBV star from detection on pre-explosion images (Gal-Yam & Leonard Reference Gal-Yam and Leonard2009). In the case of the transient SN 2009ip (more details on the environment of this intriguing event below), pre-eruption data appear to suggest a very luminous, very massive star. If very massive star progenitors were the case for all SNe IIn, then one would expect their explosion sites to be similar to the environments of very massive stars or other SNe expected to arise from such massive star populations. In Figure 18, we plot the cumulative Hα NCR distributions of SNe IIn compared to those of other SN types, as taken from Habergham et al. (Reference Habergham, Anderson, James and Lyman2014). The surprising result is observed that SNe IIn show the lowest degree of association with host galaxy Hα emission of all CC SN types. This would appear to suggest that SNe IIn have lower progenitor masses than previously assumed. A specific analysis of SN IIn environments is presented in Habergham et al. (Reference Habergham, Anderson, James and Lyman2014). In Figure 19, cumulative NCR pixel statistics are shown with respect to host galaxy near-UV emission. This shows that the SNe IIn accurately trace the recent SF within their host galaxies, in a very similar fashion to SNe IIP. The similarity of the distributions of SNe IIn and SNe IIP with respect to both Hα and near-UV emission, suggests that their progenitor lifetimes, and hence masses are similar. In addition, in Figures 18 and 19 we also plot the SN Ic population. The SN Ic distributions are clearly offset from that of SNe IIn, and if we assume that SNe Ic arise from the highest progenitor masses of CC SNe, this is further evidence that SN IIn have relatively lower mass progenitors (we also note the differences in their radial distributions seen in Habergham et al. Reference Habergham, Anderson, James and Lyman2014). Another interesting aspect of Figure 19 is that SNe Ic show a higher than one-to-one association to the near-UV emission. This is consistent with what was previously observed by Kelly et al. (Reference Kelly, Kirshner and Pahre2008) investigating SN pixel statistics with respect to host galaxy g′-band light and again implies that the most massive SF is clustered to the brightest UV regions within galaxies.

Figure 18. Cumulative Hα NCR distributions of the interacting transients SNe IIn and ‘impostors’ as compared to SNe Ia, SNe II, and SNe Ic. This figure is Figure 5 taken from Habergham et al. (Reference Habergham, Anderson, James and Lyman2014).

Figure 19. Cumulative near-UV NCR distributions of the interacting transients SNe IIn and ‘impostors’ as compared to SNe Ia, SNe II, and SNe Ic. This figure is Figure 6 taken from Habergham et al. (Reference Habergham, Anderson, James and Lyman2014).

There are a number of transients discovered where it is not clear whether they are the deaths of stars, or a large-scale eruption which produces a bright (but significantly dimmer than most SNe) transient. These transients have spectra very similar to SNe IIn, and are named SN ‘impostors’ (see van Dyk et al. Reference van Dyk, Peng, King, Filippenko, Treffers, Li and Richmond2000; Maund et al. Reference Maund2006, and Kochanek, Szczygieł, & Stanek Reference Kochanek, Szczygieł and Stanek2012 for a recent discussion). The environments of these events were also analysed and discussed in Habergham et al. (Reference Habergham, Anderson, James and Lyman2014), and their pixel statistics are shown in Figures 18 and 19. Here, we find that ‘impostors’ show an even lower degree of association to SF within galaxies than SNe IIn. However, it is important to note that the much lower luminosities of ‘impostors’ compared to other SNe means that significant selection effects could be at play in the discovery of these transients against bright H ii regions.

SN 2009ip is a transient with an unclear nature. Indeed, it is still debated whether the star has finally terminally exploded or that the transient is still being powered by eruptions and interactions (Smith et al. Reference Smith2010; Prieto et al. Reference Prieto, Brimacombe, Drake and Howerton2013; Pastorello et al. Reference Pastorello2013; Mauerhan et al. Reference Mauerhan2013; Fraser et al. Reference Fraser2013; Margutti et al. Reference Margutti2014; Smith, Mauerhan, & Prieto Reference Smith, Mauerhan and Prieto2014; Graham et al. Reference Graham2014; Mauerhan et al. Reference Mauerhan2014). One extremely interesting aspect of this transient that is often overlooked is the region within its host galaxy where it exploded, i.e. its environment. As outlined in particular in Fraser et al. (Reference Fraser2013), SN 2009ip occurred in the very outer regions of its host galaxy, more than 4 kpc from the galaxy centre, with the vast majority of the host galaxy light being contained within the isophotal ellipse that passes through the SN position (indeed, Fraser et al. Reference Fraser2013 estimated an Fr _R value of ~ 1). This is shown in Figure 20, reproduced from Fraser et al. (Reference Fraser2013). While it has been shown in Habergham et al. (Reference Habergham, Anderson, James and Lyman2014) that SNe IIn and ‘impostors’ are generally found to more frequently explode within the outer regions of hosts than e.g. SNe IIP, SN 2009ip is still at the extreme of these distributions. The explosion site of SN 2009ip is somewhat surprising given that there are strong indications that the progenitor was or is a very massive star of > 60 M_⊙ (Foley et al. Reference Foley2011). However, there is no sign of ongoing or recent SF in the immediate vicinity of SN 2009ip, and neither is there any discernible spiral arm near to the explosion location (Fraser et al. Reference Fraser2013). If the progenitor of SN 2009ip was such a massive star, as seems to be the case, then its environment is puzzling. It may be that this transient is simply an example of extreme isolated SF. It is also possible that the radial position of the explosion implies a low-metallicity environment and hence progenitor. For now, the environment of SN 2009ip remains a mystery; however, it will be intriguing whether such isolated regions within galaxies are typical of these types of events, once a statistical analysis is possible in the future.

Figure 20. The environment location of SN 2009ip within its host galaxy. This figure is a reproduction of Figure 22 from Fraser et al. (Reference Fraser2013).

3.6.4 87A-like events

Taddia et al. (Reference Taddia2013a) studied specifically the environments of SNe with similar properties to SN 1987A. SN 1987A is the closest SN observed in the modern era of astronomy. SN 1987A was peculiar in that its light curve showed a slow extended rise (see e.g. Arnett et al. Reference Arnett, Bahcall, Kirshner and Woosley1989). There are now a handful of similar SNe which have been discovered and studied (Pastorello et al. Reference Pastorello2005; Kleiser et al. Reference Kleiser2011; Pastorello et al. Reference Pastorello2012; Taddia et al. Reference Taddia2012; Arcavi et al. Reference Arcavi2012). SN 1987A itself exploded in the Large Magellanic Cloud (LMC) at a distance of only 50 kpc. The wealth of information from pre-explosion observations enabled detailed studies of its progenitor star which was observed to be a blue supergiant star. Given the unusual nature of this SN, it is interesting to ask whether these peculiarities are related to the specific environment within which it exploded.

Taddia et al. (Reference Taddia2013a) studied the environments of 13 87A-like SNe using very similar methods to those outlined above. Figure 21, from Taddia et al. (Reference Taddia2013a), shows the cumulative distributions of the 87A-like population with respect to other CC SNe in terms of: host galaxy absolute magnitudes; galactocentric radial positions; and host H ii region metallicities. It appears that 87A-like SNe prefer environments with somewhat lower metallicities than other CC SNe, as indicated directly by the measured oxygen abundances of their host H ii regions, and confirmed by their fainter host galaxy luminosities and higher radial distances within these hosts. Taddia et al. also investigated NCR values for their 87A-like sample, finding a similar distribution to SNe IIP.

Figure 21. Comparison of the environments of 87A-like events with other CC SNe. Left: cumulative host galaxy absolute magnitude distributions for CC SNe. Middle: cumulative R ₂₅ normalised galactocentric radial distributions of CC SNe. Right: cumulative host H ii region oxygen abundance distributions for CC SNe. This figure is of Figure 1 from Taddia et al. (Reference Taddia2013a) and is reproduced by permission of A&A.

3.7 Correlating specific transient features with environment properties

To date, environment and host galaxy studies have generally concentrated on investigating differences in the environmental properties of SNe when events are simply split by spectroscopic type. These typings are often somewhat subjective, and in the case of SNe IIP and SNe IIL, spectroscopic typing without any further light-curve information used can simply be wrong. This is an issue with SN catalogues which are often not updated after initial classifications are given in circulars. Indeed, it has been shown in Gutiérrez et al. (Reference Gutiérrez2014) that while correlations between SN II spectral parameters and light-curve morphologies do exist, there is significant dispersion, which can lead to ambiguous spectroscopic classifications. In the case of SNe IIb and SNe Ib, the epoch at which the classification spectrum is taken can then lead to an event being typed into one class or the other (see e.g. Milisavljevic et al. Reference Milisavljevic2013 for discussion on this issue). Given these ambiguities, it makes sense for future statistical studies to use more specific SN features to investigate environment trends. While historically this type of analysis was difficult due to the lack of large SN follow-up databases, recent years have seen a number of publications of statistically significant samples of CC SNe (SN Ia large samples have been relatively more widespread, as is seen in the analysis in Section 3.5). SN II photometry, spectra and measured parameters for large samples have been recently published in Arcavi et al. (Reference Arcavi2012), Anderson et al. (Reference Anderson2014a, Reference Anderson2014b), Gutiérrez et al. (Reference Gutiérrez2014), Faran et al. (Reference Faran2014b), Sanders et al. (Reference Sanders2015), and Faran et al. (Reference Faran2014a). SNe Ibc and SNe IIb samples have been published in Drout et al. (Reference Drout2011), Bianco et al. (Reference Bianco2014), and Modjaz et al. (Reference Modjaz2014). This provides an opportunity to correlate environment properties to more specific features such as light-curve morphologies, the presence and form of specific spectral features and their evolution in time, and parameters that have a direct link to explosion and progenitor properties such as synthesised nickel mass.

In this section, we analyse the environmental properties of the SN II sample presented and analysed in Anderson et al. (Reference Anderson2014b). That work presented a sample of more than 100 well sampled V-band light-curves from which key parameters were defined then measured. A full analysis of all parameters and all possible environment measurements will be left for future publications. Here, we present a few interesting figures using the analysis methods outlined in previous sections. In Anderson et al. (Reference Anderson2014b), a parameter ‘s ₂’ was defined to give the decline rate of SNe II during their ‘plateau’ phases (which are not necessarily flat, see that publication for in-depth discussion). Using such a parameter allows one to move away from simple ‘IIP’ or ‘IIL’ classifications. In Figure 22, we present near-UV NCR distributions for SNe II when the sample is split into low and high s ₂ populations (slow and fast decliners). One can see that the faster decliners show a higher degree of association to the recent SF within galaxies.

Figure 22. Near-UV NCR distributions for SNe II when these SNe are split by their light-curve decline rate parameter: s ₂. The latter values are taken from Anderson et al. (Reference Anderson2014b).

Host galaxy SN extinction is a key parameter in obtaining intrinsic SN brightnesses, however especially in the case of SNe II, a clear method for estimating this parameter is not forthcoming (see e.g. discussion in Faran et al. Reference Faran2014b). One measurement which is often used to define the presence of host galaxy A_v and its amplitude is the EW of narrow sodium lines within spectra. However, the accuracy of this measurement is debated (see discussion in Phillips et al. Reference Phillips2013 as applied to SNe Ia). It can be thus enlightening to see how sodium absorption within SNe II spectra changes as a function of environment. In Figure 23, Fr _R distributions are plotted for SNe II which do/do not show the presence of narrow sodium absorption within their spectra. One can see that those events with detections occur at smaller galactocentric radii than those without. This is to be expected as within the more central regions within galaxies a higher degree of ISM extinction is expected.

Figure 23. Fr _R distributions for SNe II with and without narrow sodium absorption detections within their spectra.

Finally, in Figure 24 we show SN II H ii region metallicities, when the sample is split by the Anderson et al. (Reference Anderson2014b) parameter OPTd which is a measure of the optically thick phase duration of the light-curve and measures the time between explosion and the end of the ‘plateau’ stage. It is generally assumed that the length of this light-curve phase is related to the mass of the hydrogen envelope at the epoch of explosion. While the statistics are small, it is observed that SNe with longer OPTd phases appear to explode in higher oxygen abundance regions than their shorter OPTd counterparts.

Figure 24. H ii region metallicities for SNe II when split into equal samples of OPTd, the optically thick phase durations of their light-curves. The latter values are taken from Anderson et al. (Reference Anderson2014b).

5.1 New types of environment observations

We have already commented on the central importance of wide-field IFUs for future studies of SN host galaxies. This enables simultaneous spatial and spectral information over the entire extent of host galaxies. Indeed, the usefulness of IFU observations for SN host environment studies has already been demonstrated by e.g. Stanishev et al. (Reference Stanishev, Rodrigues, Mourão and Flores2012), Rigault et al. (Reference Rigault2013), Kuncarayakti et al. (Reference Kuncarayakti2013a, Reference Kuncarayakti2013b) and Galbany et al. (Reference Galbany2014).

MUSE mounted on the VLT telescope promises to significantly enhance the capabilities of IFU instrumentation for environment studies, having a FOV of 1 arcmin × 1 arcmin, extremely good spatial resolution (0.2 arcsec spaxel size), and the power of being attached to an 8-m class telescope. We recently obtained Science Verification data (P.I. Galbany), to observe the galaxy NGC 6754 which has been host to four SNe. A full analysis of that galaxy and other data taken during SV in terms of SN environmental parameters will be presented in L. Galbany et al. (in preparation). Here, we show a glimpse of the data obtained and show what can be achieved through such analysis. (Note, an analysis of the H ii region population of NGC 6754 using these data has been presented in Sánchez et al. Reference Sánchez2015.) In Figure 25, we present the spatial metallicity map of NGC 6754 on the PP04 O3N2 scale, with SN positions indicated. In Figure 26, a histogram of all measured metallicities is shown, with values estimated at exact SN sites indicated. These figures show the power of these data as compared to previous work, much of which was discussed above, where in the latter either spatial information of a specific emission line (e.g. Hα) has been analysed, or spectral information but in one small-scale region. With current state-of-the-art instrumentation such as MUSE, one can obtain all these parameters with one single telescope pointing.

Figure 25. PP04 O3N2 metallicity map of NGC6754, with SN explosion sites indicated.

Figure 26. Histogram of PP04 O3N2 metallicity measurements of NGC 6754, with SN environment metallicity positions indicated.

Additional instruments are also coming online to further these IFU possibilities, such as SAMI (Bryant et al. Reference Bryant, McLean, Ramsay and Takami2012), and MANGA (Law & MaNGA Team Reference Law2014). WiFeS (Dopita et al. Reference Dopita2010; Childress et al. Reference Childress, Vogt, Nielsen and Sharp2014a) is another instrument which is being used for these types of studies. These multi-IFU instruments will observe thousands of galaxies. While SN environment science may not initially be the key science drivers, when one observes such quantities of galaxies, then one naturally obtains statistically significant samples of SN host galaxies, which can be used for environment studies. Using these instruments, one can essentially analyse host galaxies/environments in the manner described in Sections 2.1–2.3simultaneously, i.e. one can extract any emission line (obviously within the constraints of any given instrument), or broad-band ‘colour’ and proceed with ‘spaxel statistics’. This will enable environment parameters such as metallicity to not only be compared between SN types, but also between SN environments and all other environments within their host galaxy (this in essence is what is done above for the Hα pixel statistics). This is what is shown in Figure 26: a histogram of all host galaxy environment metallicity measurements. In this particular galaxy, it is observed that the SNe Ia explode within the peak of the metallicity distribution, while the SNe II appear to explode in regions of higher oxygen abundance. We note here that this analysis is very preliminary, and is shown here to present future possibilities. Published values may change once a more thorough analysis is achieved.

While large-scale statistical environment analysis will hugely benefit with the advent of wide-field IFU instrumentation, it is also interesting to attempt to observe environments with the best spatial resolution in order to observe individual progenitor parent stellar clusters. In order to obtain the best spatial resolution, adaptive optics (AO) helps to overcome the limitations of atmospheric conditions (i.e. seeing) during an observation. In this fashion, we have obtained data of a handful of SN environments with SINFONI (Eisenhauer et al. Reference Eisenhauer, Iye and Moorwood2003), a near-IR IFU instrument. In Figure 27, we present an example of the gain in spatial resolution using near-IR AO observations, where SINFONI observations are compared to those of an optical IFU of the same environment. This shows that what appears to be a single cluster in an optical observation is resolved into several clusters in SINFONI observations. In the near-IR, one can use the presence or absence of Br-gamma emission to constrain the age of the environment/parent cluster (more details will be presented in H. Kuncarayakti et al. in preparation). Better spatial resolution also enables us to alleviate SF history issues.

Figure 27. Example of the advantages of moving to the near-IR where one can obtain much higher spatial resolution due to adaptive optics corrections. A comparison is made between optical IFU (SNIFS) observations, and those obtained with SINFONI of the same SN environment. Also shown are the extracted near-IR spectra, where the Br-gamma emission line is indicated. (A similar figure is also shown in Kuncarayakti et al. Reference Kuncarayakti2014)

The above examples show how the field of environments is set to evolve over the coming years. In addition, one assumes that investigators will find other innovative ideas to use other instruments to study SN environments. In combination with growing databases of high-quality follow-up data, the coming years promise to be particularly fruitful for the environment and SN fields.