Chemical enrichment involves supernova explosions, stellar winds, and planetary nebula ejections. Events involving massive stars are concentrated in young star clusters, leading to the potential for localized enrichment. Infall of primordial (or not so primordial) gas from the environment of a galaxy may involve accretion of gas clouds or dwarf galaxies, possibly leading to localized depressions of heavy element abundances. Heavy elements are dispersed by flows propelled by supernova explosions and stellar winds, as well as differential rotation, large scale flows induced by bars, etc.
Local abundance fluctuations in spiral galaxies have a smaller amplitude than the overall radial gradients. There is, however, evidence for significant fluctuations in the Milky Way. The solar oxygen abundance [O/H] = 8.87 (Grevesse & Noels 1993) significantly exceeds the Orion nebula value 8.58 (Baldwin et al. 1991; Esteban et al. 1998) as well as HW's ``composite'' H II region value 8.68 ± 0.05 at the solar circle. (One uncertainty is the nagging question of temperature fluctuations in the nebular gas [Peimbert 1967]). Abundances in B stars in the vicinity of the Orion complex agree with the nebular value, although there is evidence that abundances increased as star formation progressed through the region (Cunha & Lambert 1992). The sun is 0.17 dex more metal rich than the average nearby star (Wielen, Fuchs, & Dettbarn 1996). Wielen et al. propose that the sun was born ~ 2 kpc closer to the Galactic center than its current orbital radius, where abundances were higher. Dispersions among open clusters presumably are less affected by this process, if indeed it is important (Garnett & Kobulnicky 2000).
Edvardsson et al. (1993) found a large spread in [Fe/H] in G stars at the solar circle, but little trend in metallicity with age. Friel & Boesgaard (1992) studied C and Fe abundances in F stars in open clusters in the Galactic disk. They found no significant dispersion in abundance among the stars in a single cluster at the level of 0.05 dex but significant differences from cluster to cluster at a level of ~ 0.1 dex. They found little dependence on age from 50 million to 5 billion years. The open clusters studied by Carraro et al. (1998) show a range in [Fe/H] of 0.6 dex, suggesting a dispersion 0.15. In contrast, interstellar oxygen measurements by Meyer, Jura, & Cardelli (1998) show a dispersion of only ± 0.05 dex on a number of sightlines. In a recent study of the Galactic H II regions by Deharveng et al. (2000), the scatter of the best measurements around the mean gradient in O/H is small. Garnett & Kobulnicky (2000) examine evidence for abundance fluctuations in H II regions, stars, and the interstellar medium. They argue that selection effects in the star sample of Edvardsson et al. complicate the interpretation of the abundance scatter, and that the true disperson for field stars is less than 0.15 dex.
Results for external spirals are inconclusive, because of the lack of precise measurements of the electron temperature in a sufficient sample of GEHRs. In a study of H II regions in M11, Kennicutt & Garnett (1996) suggest, on the basis of line ratio correlations, that the true dispersion around the mean gradient is less than the observed scatter of 0.1 to 0.2 dex.
These results suggest that, in the disks of spiral galaxies, abundance fluctuations with 0.1 dex occur on a scale larger than the clouds that form a typical open cluster, but smaller than the distance between open clusters. The dominant cause of the fluctuations remains unclear. Several processes have been discussed, including local enrichment by supernovae and accretion of clouds of metal poor gas (e.g., Carraro et al. 1998, and references therein). Franco et al. (1988) discuss the possibility that the Orion molecular cloud complex, with its relatively low metallicity, results from the impact of a metal poor gas cloud.
The prevailing level of abundance fluctuations in the ISM depends on the competition between localized enrichment and mixing. Roy & Kunth (1995) considered mixing on different scales in the Galactic disk. On scales of 1 to 10 kpc, local abundance fluctuations are smoothed in less that 109 yr by turbulent diffusion of clouds together with differential rotation. On scales 100 to 1000 pc, cloud collisions, star-formation driven flows, and differential rotation mix the gas in 108 yr. On scales 1 to 100 pc, turbulence in ionized regions mixes the gas in 106.3 yr or less.
Abundances in irregular galaxies generally are quite uniform. Kobulnicky & Skillman's (1997) results for NGC 1569 show a dispersion of ± 0.05 dex in O/H and N/O in the well measured regions. They argue that, at the low metallicity of this object, pollution by the ejecta of only a few massive stars should give measurable enrichment. They suggest that the supernova ejecta form a hot wind that escapes the galaxy, disperses the ejecta, and later reaccretes, avoiding local enrichment at the site of the stars' deaths. There are, however, rare cases of local enrichment, in particular an H II region with high N/O in NGC 5253 (Kobulnicky et al. 1997).