0706.0238/ms.tex
1: \documentclass{emulateapj}
2: %\documentclass[12pt,preprint]{aastex}
3: %\documentclass[manuscript]{aastex}
4: \usepackage{natbib}
5: \usepackage{epsfig}
6: \begin{document}
7: 
8: 
9: %\slugcomment{Submitted for publication in the Special GALEX ApJ. Suppl. Issue}
10: %\slugcomment{Accepted for publication in ApJ}
11: 
12: \title{Nitrogen Production in Starburst Galaxies Detected by {\sl GALEX}}
13: 
14: \author{ Ryan P. Mallery\altaffilmark{1}, 
15: Lisa Kewley\altaffilmark{2},
16:  R. Michael Rich\altaffilmark{1}, 
17: Samir Salim\altaffilmark{1}, 
18: Stephane Charlot\altaffilmark{3,4},
19:  Christy Tremonti\altaffilmark{5},
20:   Mark Seibert\altaffilmark{6}, 
21: Todd Small\altaffilmark{6}, 
22: Ted Wyder\altaffilmark{6}, 
23: Tom A. Barlow\altaffilmark{6},
24: Karl Forster\altaffilmark{6},
25: Peter G. Friedman\altaffilmark{6},
26: D. Christopher Martin\altaffilmark{6},
27: Patrick Morrissey\altaffilmark{6},
28: Susan G. Neff\altaffilmark{7},
29: David Schiminovich\altaffilmark{8},
30: Luciana Bianchi\altaffilmark{9},
31: Jose Donas\altaffilmark{11},
32: Timothy M. Heckman\altaffilmark{12},
33: Young-Wook Lee\altaffilmark{10},
34: Barry F. Madore\altaffilmark{14},
35: Bruno Milliard\altaffilmark{11},
36: Alex S. Szalay\altaffilmark{12},
37: Barry Y. Welsh\altaffilmark{13}, 
38: Suk Young Yi\altaffilmark{10}}
39: 
40: \altaffiltext{1}{Department of Physics and Astronomy, University of California, Los Angeles, CA 90095-1562}
41: \altaffiltext{2}{Institute for Astronomy, University of Hawaii, Honolulu, HI, 4234092309239805} 
42: \altaffiltext{3}{Max-Planck Institut f\"ur Astrophysik, D-85748 Garching, Germany}
43: \altaffiltext{4}{Institut d'Astrophysique de Paris, CNRS, 98 bis boulevard Arago, F-75014 Paris, France}
44: \altaffiltext{5}{Steward Observatory, 933 North Cherry Avenue, Tucson, AZ 85721}
45: \altaffiltext{6}{California Institute of Technology,MC 405-47, 1200 East California Boulevard, Pasadena, CA 91125}
46: 
47: \altaffiltext{7}{Laboratory for Astronomy and Solar Physics, NASA Goddard
48: Space Flight Center, Greenbelt, MD 20771}
49: 
50: \altaffiltext{8}{Department of Astronomy, Columbia University, New York, NY 10027}
51: 
52: \altaffiltext{9}{Center for Astrophysical Sciences, The Johns Hopkins
53: University, 3400 N. Charles St., Baltimore, MD 21218}
54: 
55: \altaffiltext{10}{Center for Space Astrophysics, Yonsei University, Seoul
56: 120-749, Korea}
57: 
58: \altaffiltext{11}{Laboratoire d'Astrophysique de Marseille, BP 8, Traverse
59: du Siphon, 13376 Marseille Cedex 12, France}
60: 
61: \altaffiltext{12}{Department of Physics and Astronomy, The Johns Hopkins
62: University, Homewood Campus, Baltimore, MD 21218}
63: 
64: \altaffiltext{13}{Space Sciences Laboratory, University of California at
65: Berkeley, 601 Campbell Hall, Berkeley, CA 94720}
66: 
67: \altaffiltext{14}{Observatories of the Carnegie Institution of Washington,
68: 813 Santa Barbara St., Pasadena, CA 91101}
69: 
70: 
71: 
72: 
73: \begin{abstract}
74: We investigate the production of nitrogen in star forming galaxies with ultraviolet 
75: (UV) radiation detected by the Galaxy Evolution Explorer Satellite ({\sl GALEX}). We use a 
76: sample of 8,745 {\sl GALEX} emission line galaxies matched to the Sloan Digital Sky 
77: Survey (SDSS) spectroscopic sample. We derive both gas-phase 
78: oxygen and nitrogen abundances for the sample, and apply stellar population synthesis models 
79: to derive stellar masses and star formation histories of the galaxies.
80:  We compare oxygen abundances derived using three different diagnostics. 
81:  We derive the specific star formation rates 
82:  of the galaxies by modeling the $7-$band {\sl GALEX}$+$SDSS photometry.
83: We find that galaxies that have log SFR/M$_*\gtrsim-10.0$ typically have values of log N/O $\sim0.05$ dex less than 
84: galaxies with  log SFR/M$_* \lesssim-10.0$ and similar oxygen abundances. 
85: \end{abstract}
86: \keywords{galaxies: abundances - galaxies: fundamental
87:   parameters - galaxies: starburst - ultraviolet: galaxies}
88: 
89: 
90: 
91: 
92: \section{Introduction}
93:            The abundance of nitrogen in
94:            galaxies and the site of its creation is critical
95:            for our understanding of galaxy chemical evolution.
96:            The ratio of N/O is especially useful because both of these elements are
97:            created by different mechanisms in different ranges of stellar mass.
98: %While the abundance of metals depends on the initial mass function (IMF), and the star formation  history,
99: % the abundance of nitrogen in particular has a dependence on the metallicity \citep{garnetbook}. The main questions are whether 
100: %it is completely  dependent on the initial metallicity of the star and which populations are  responsible for its production.
101:            Nitrogen is produced during hydrogen burning via the CNO
102:            and CN cycles, and is created as both a primary and secondary element.
103:            In primary nucleosynthesis the production of nitrogen is
104:            independent of the initial metallicity of the
105:            star. Primary
106:            production of nitrogen occurs predominantly only
107:            in intermediate mass stars
108:            (4$\leq$M$/$M$_{\odot}$ $\leq$8)
109:            \citep{mat85, mat86} 
110:            where the nitrogen is produced during hydrogen
111:            shell burning while carbon and oxygen, which are
112:            assumed to be primary nucleosynthesis elements, are moved from
113:            the core to the outer stellar layers during
114:            dredge-up episodes. Recent stellar models that include
115:            rotational effects indicate that massive
116:            stars between 9 and 20 M$_{\odot}$ may produce primary nitrogen \citep{maeder}. These massive star models
117:            incorporate a convective helium burning shell that penetrates into the hydrogen
118:            burning shell, creating primary
119:            nitrogen. The discovery of metal-poor halo stars with high N/O ratios by \citet{spite} and \citet{israelian}
120:            seems to confirm the primary production of nitrogen in massive stars with a yield that depends on stellar mass and metallicity \citep{chia05, chia06}. 
121:            In secondary production, nitrogen is synthesized
122:            from the carbon and oxygen initially present in the
123:            star and its abundance is therefore proportional to
124:            the initial heavy element abundance.  Secondary production is common to
125:            stars of all masses \citep{mat85,mat86}. Though the synthesis of nitrogen in stars is becoming better understood, 
126:            our understanding of the abundance of 
127:            nitrogen in galaxies is lacking. One way to investigate  the primary versus secondary origin
128:            of nitrogen is to examine the ratio of N/O as a function of O/H. These abundance ratios are commonly computed from 
129:            optical nebular emission lines of HII regions (see \S 5). In the case of primary nucleosynthesis N/O will be constant; secondary enrichment 
130:            produces a linear correlation between log N/O and log O/H. The combination of primary and secondary nucleosynthesis produces a 
131:            non-linear relation (see Fig. 3). The abundance ratios of many other elements such as neon, sulfur, and iron  with respect to oxygen have been found to 
132:           tightly correlate with O/H \citep{iz05}, yet at fixed O/H galaxies have been found to have a scatter in N/O of a factor of 2 \citep{villa93}. 
133:           Until recently, this problem could not be 
134:            adequately addressed due to small sample sizes, and uncertain abundances. 
135:                      
136:            Chemical evolution scenarios proposed to explain this scatter include (1) a primary plus
137:            secondary origin of nitrogen but with variable
138:            initial mass functions (IMFs) \citep{alloin},
139:            (2) a primary plus secondary origin but with a time delay
140:            between the release of nitrogen and the release of oxygen back into
141:            the interstellar medium (ISM) \citep{villa93, garnet90, thurston}. With
142:            regards to the former scenario,
143:            \citet{chia} have found that an IMF constant in
144:            space and time better reproduces the observational
145:            constraints of the solar neighborhood  (i.e.
146:            the ratio of metal-poor to metal-rich stars, the ratio of SN II to SN I,
147:            and the ratio of He to metal abundance.). \citet{chia} find that such an IMF also helps reproduce the observed abundance
148:            gradient of the Galactic disk more reliably than models with IMFs that depend on metallicity or SFR. 
149:             In the time delay chemical evolution model,
150:            oxygen is released in the supernovae
151:            of short lived massive stars. As the metallicity of successive generations of  massive stars increases,
152:            secondary nitrogen is also released, and then the bulk of nitrogen is
153:            released much later in intermediate mass stars. The chemical evolution models of \citet{field}  
154:             for blue compact galaxies and \citet{chia03} for dwarf galaxies, 
155:            incorporating the variation of stellar yields with stellar mass and stellar evolution timescales, 
156:            have shown that the scatter in N/O could be reproduced by varying star formation histories.
157: 
158:                               
159:            Work by \citet{considere} on
160:            abundances in barred spiral
161:            galaxies indicates that the nitrogen abundance
162:            is the result of both primary and secondary
163:            nucleosynthesis. However, nitrogen abundances taken from a small UV selected
164:            galaxy sample detected by FOCA\footnote{FOCA was a
165:            balloon-borne 40 cm telescope that imaged at
166:            2015\AA, FWHM 188\AA~\citep{mil}.} \citep{contini} show mostly a secondary component, 
167:            but still with considerable scatter. They proposed that the difference between the two results arises because there is a time delay
168:            between the release of oxygen and nitrogen. The UV galaxies were starbursts;
169:           consequently, the high mass stars formed during the starburst had released newly synthesized oxygen into the ISM.
170:            This increased the oxygen abundance 
171:            and lowered the N/O  ratio of these galaxies such that their new nitrogen and oxygen 
172:            abundances were consistent with only a secondary nitrogen component. Then later, 
173:               once the intermediate mass stars formed during the burst have had time to
174:            evolve, nitrogen will be released into the ISM. This would increase the nitrogen in these UV galaxies to a similar  amount of N/O
175:            as the sample of \citet{considere}.
176:          
177:              The time delay scenario, as stated  by \citet{contini}, may be oversimplified. It does not account for 
178:              galaxies that have  star formation histories differing from cycles of bursting  phases followed by 
179:              quiescent phases, and was made prior to any evidence for the production of primary 
180:            nitrogen in massive stars \citep{maeder, spite, chia05,chia06}. A more accurate statement is: the release of material by 
181:            a star of mass, M into the ISM will cause the N/O ratio of the ISM  to increase if its relative yield of p$_N$/p$_O$ is greater than 1 and decrease
182:            if its relative yield  is less than 1. The new nitrogen and oxygen stellar yields of \citet{chia06}  
183:            still indicate that while massive stars (M$_* >9$M$_{\odot}$) produce primary nitrogen, the ratio of nitrogen to oxygen yields for massive stars
184:             is $\lesssim 1$ and still decreases with increasing mass. Thus, a burst of  star formation will 
185:             still initially cause a decrease in N/O, 
186:             the effect of the new yields only diminishes the extent to which a starburst can conceivably lower N/O. 
187:                All that is required for the N/O  of a galaxy to increase 
188:             is that its current SFR is less than its past average SFR such that comparatively fewer high mass stars are being formed, allowing 
189:             for intermediate mass stars of previous generations to 
190:             dominate the chemical enrichment of the galaxy.
191:  
192:   
193:            %add something about the large number of SDSS galaxies and its advantages
194:            The large amount of uniformly calibrated data from the Sloan Digital Sky Survey (SDSS) has recently enabled
195:            robust statistical studies of chemical  enrichment \citep{trem05}.
196:            Recently, \citet{iz05} examined the ratio of N/O  in metal-poor galaxies, 
197:            $12+\log O/H <8.5$, in SDSS  DR3 and
198:            concluded that the N/O ratio increases with increasing starburst age (decreasing EW$_{H\beta}$) 
199:            for metal-poor galaxies, due to the ejection of 
200:            nitrogen by Wolf-Rayet stars. \citet{liang}
201:            consider a  $\sim 30,000$ galaxy SDSS DR2 sample, and show that objects with higher N/O tend to have lower $EW_{H\alpha}$.
202:         This result is consistent with  those galaxies 
203:             with current star formation rates that are high with respect to their past average star formation rate, exhibiting a higher
204:            oxygen abundance. The oxygen is presumably contributed by the recently formed massive stars.
205:             Here we examine the N/O ratio as a function of
206:             O/H in a sample of UV selected galaxies
207:            detected by the Galaxy Evolution Explorer (GALEX),
208:            exploiting the overlap between the {\sl GALEX} Medium Imaging Survey and the SDSS
209:            spectroscopic footprint in the local
210:            universe ($z<0.3$). The large volume of data available from {\sl GALEX}+SDSS
211:           makes it possible to go beyond equivalent widths and calculate physical quantities like stellar masses, M$_*$, and specific
212:           star formation rates, SFR/M$_*$, which can be more easily compared with models when investigating chemical enrichment.
213:               We use the SFR/M$_*$ and  $M_*$, of these galaxies
214:            derived from their 7-band UV-optical photometry \citep{ss07} to test whether the star formation history
215:            of a galaxy can explain the
216:            observed relationship between the nitrogen and oxygen abundances in our sample and 
217:             investigate the accuracy of determined abundances.  
218:          We note that O/H is not equivalent to a time axis, and the values of  N/O and  O/H 
219:          represent the current chemical evolutionary stages for galaxies that 
220:          have most likely had different histories of star formation and other dynamical processes such as galactic winds
221:          and gas accretion timescales \citep{chia03, diaz86, mat85}. We expect, nevertheless, that if the time delay scenario
222:          is correct, that galaxies currently exhibiting a 
223:          strong burst of star formation will on average have lower values of N/O than non-bursting galaxies at similar metallicities.
224: 
225: 
226:           The outline of the paper is as follows. In
227:           \S 2 we describe the data and sources used in this
228:           analysis. We present our galaxy
229:           sample containing matched GALEX and SDSS sources in
230:           \S 3. An explanation of the derivation of galaxy
231:           parameters by matching the models of \citet{bc03} to
232:           the 7 color UV-optical SED of each source is
233:           given in \S 4. We describe and contrast the various
234:           methods used in determining oxygen
235:           abundances in \S 5 and nitrogen in \S 6. In
236:           \S 7  we examine the relationship of nitrogen to oxygen for our sample  and
237:           in \S 8 we give our conclusions. We assume {\it H$_o$}=70
238:           km s$^{-1}$ Mpc$^{-1}$, $\Omega_{m}$=0.3, and  $\Omega_{\Lambda}$=0.7.
239:  
240:      
241: \section{Data}
242:    We consider galaxies with GALEX photometry from the Medium
243:   Imaging Survey (MIS) Internal Release 1.1 (IR1.1), m$_{lim}$(AB)$\approx$ 23, and SDSS
244:   photometry and spectra.  {\sl GALEX}  is a NASA Small
245:   Explorer Mission that aims to survey the UV emission from Galactic and extragalactic sources
246:   from a 700km circular orbit
247:   \citep{martin,pm}. {\sl GALEX}  images the sky simultaneously in two bands, 
248:   the far-UV (FUV 1344-1786\AA) and the near-UV (NUV 1771-2831
249:   \AA). Each {\sl GALEX}  circular field is 1.25 deg. in  diameter. We use
250:   FUV and NUV magnitudes and magnitude errors
251:   derived in elliptical apertures\footnote{{\sl GALEX}  source
252:   detection and measurement is obtained from SExtractor
253:   \citep{bar}}. 
254: 
255:    We use optical photometry for our objects obtained from
256:    SDSS Data Release 4 (DR4) spectroscopic sample. Most of our objects were taken from
257:    the main galaxy spectroscopic survey ({\it r}$_{lim}<$17.8), but many of our objects were originally 
258:    targeted as quasars and taken from the quasar spectroscopic survey  ({\it r}$_{lim}<$19.5 \citet{york}). %%%%%%%%%check thiss
259:    The SDSS photometric data are taken with
260:   the 2.5m telescope at Apache Point Observatory. Imaging is
261:   obtained in  {\it ugriz} bands
262:   \citep{fuk96,smith02}. The imaging data are photometrically
263:   \citep{hogg} and astrometrically \citep{pier} calibrated. An
264:   overview of the SDSS data pipelines and products can be
265:   found in \citet{stoughton}. 
266:   
267:  The SDSS spectra are acquired using  3\arcsec~diameter fibers
268:   positioned on the centers of the target galaxies. The spectra are flux and wavelength calibrated for
269:   wavelengths between 3800-9200\AA~at resolving power \citep{york}
270:   R$\equiv$$\lambda$$/$$\Delta$$\lambda$ =1850-2200. 
271:   We use continuum subtracted emission-line fluxes and flux errors from the
272:   SDSS spectra measured by \citet{trem05}, to divide and classify the
273:   sample in terms of emission-line ratios, and to derive
274:   nebular abundances.  
275: 
276: \section{The Sample}
277:    We use a sample constructed by matching objects with {\sl GALEX} MIS IR1.1
278:    detections to galaxies in the SDSS DR4 spectroscopic sample. The objects are matched 
279: within a 4\arcsec~radius \citep{sbrt05,ss07}. We accept only
280: unique matches and discard objects that contain multiple matches. 
281: We restrict the sample to galaxies with z$<0.3$ 
282: that are detected by {\sl GALEX} in the NUV at a 3$\sigma$ level.
283: We further restrict the sample to galaxies with spectral {\it r}-band fluxes greater
284: than  20\% of their total {\it r}-band fluxes. \citet{kjg} found that in samples where
285: the spectroscopic fiber collects greater than 20\% of the
286: galaxy light, the fiber metallicities approximate global values. This criterion gives a sample of 36225 objects. 
287: 
288:          In order to constrain the errors on the derived abundances for objects, we impose detection
289:          criteria for several emission lines. We remove galaxies
290:          with $<5\sigma$ detections of the Balmer lines H$\alpha$ and H$\beta$
291:          and [NII]$\lambda$6584.
292:          For other oxygen forbidden
293:          lines that are used in the analysis, [OII]$\lambda$3726,3729 [OIII]$\lambda$5007, 
294:          we remove galaxies with $<3\sigma$ detections. We note that
295:          demanding a [OII]
296:          detection restricts the galaxy's redshift
297:          to $z>0.03$ due to the wavelength cutoff of
298:          the SDSS spectrograph at 3800\AA. These constraints
299:          trim the sample to 12213 galaxies.
300:          
301:          We identify Active Galactic Nuclei (AGN) in our  sample 
302:           by using  the line diagnostic diagram
303:          [NII]$/$H$\alpha$ versus [OIII]$/$H$\beta$ \citep{bpt}.
304:          We use the formula of
305:          \citet{kauf03b} to remove galaxies with contributions to their emission line spectrum from AGN.
306:          The fraction of galaxies removed
307:          because of possible contamination due to AGN is $\sim$21\%.
308:          
309:        Other sources of emission-line flux besides star forming regions include
310:        planetary nebulae (PN), and supernova remnants (SNR). Studies by \citet{oey} of the Large
311:        Magellanic Cloud (LMC)  show that SNRs affect the
312:        emission-line spectra at a fairly low level. As
313:        discussed in \citet{cl01} the radiation from planetary
314:        nebulae can be neglected since the ionizing radiation is
315:        typically less than 0.1 percent of that produced by
316:        massive stars at an earlier age. 
317: 
318:          We finally remove 831 galaxies with failed fits to the photometric 7-band SED that give reduced $\chi^2 >10$.   
319:          This gives a sample of 8,745 galaxies, from which 72\% are from the SDSS main galaxy sample.
320:           Galaxies in the final star forming sample with 3$\sigma$ FUV
321:          detections comprise $\sim$84\% of the sample.  The emission-line criteria
322:          we use selects galaxies that are blue in NUV$-{\it
323:          r}$, with $NUV-r < 4$. {\sl GALEX} is remarkably sensitive to  star-forming galaxies. In all of the {\sl GALEX} MIS 
324:          IR1.1 fields only 155 galaxies that are detected in SDSS DR4 and that satisfy our emission line criteria do not have
325:          3$\sigma$ NUV or FUV detections. The percent {\sl GALEX} detection is $99.4$\% for  ${\it r}<17.8$ and $97.9$\% 
326:          for ${\it r}>17.8$.
327:          
328: 
329: 
330: \section{Derived Galaxy Parameters}     %DRSFT DONE
331:    We use the following galaxy parameters derived by
332:    \citet{ss07}: the NUV and FUV dust attenuations, A$_{NUV}$
333:    and A$_{FUV}$ in magnitudes, the current star formation
334:    rate, SFR, averaged over the past 100 Myr in M$_{\odot}$ yr$^{-1}$, the
335:    present-day stellar mass, M$_*$, of the galaxy in
336:    M$_{\odot}$, the specific star formation rate, SFR/M$_*$, and the fraction of stellar mass
337:   formed in starbursts over the last 100Myr, F$_{Burst}$. 
338:  Galaxy parameters are derived from model libraries of
339:  galaxies at redshifts of .05, .10, .15, .20, and .25. Each
340:  library consists of $\sim$10$^5$ models. Each model is
341:  defined by several parameters: galaxy age, optical depth, star formation history, and metallicity. 
342:  The star formation history of each model follows the prescription of  \citet{kauf03a} and consists of an underlying, continuous, exponentially
343:  declining SFR upon which bursts of star formation, random in time and amplitude, are superimposed. 
344:  Dust attenuation in each model is parametrized using the
345:  prescription of \citet{cf00}. A description of the prior distributions of
346:  the model parameters is discussed in \citet{ss05} and \citet{ss07}.
347:  
348:   Model spectral energy distributions (SEDs) are created for each set of model parameters
349:   using the population synthesis code of \citet{bc03} and
350:   assuming a \citet{kroupa} IMF.    
351:   The model SEDs are convolved with the GALEX and SDSS filter
352:   response curves. Statistical estimates of physical galaxy parameters  are derived by
353:   comparing the observed 7 band GALEX$/$SDSS  fluxes of each
354:   galaxy to all the convolved model SEDs in the nearest redshift
355:   library. Probability density functions  (PDFs) for each
356:   physical parameter are created by assigning weights to
357:   the parameters of a model. The $\chi^{2}$ goodness of fit of each
358:   model determines the weight ($\propto \exp[-\chi^{2}/2]$)
359:   that is assigned to the parameters of that model. The median
360:   of the PDF is taken as the estimate of the galaxy parameter.
361:   An estimate of the error for the parameters is taken as
362:   $1/4$ of the  2.5-97.5 percentile range. Table 1 lists the
363:   parameters and their mean errors. 
364: 
365:   
366: 
367:   %TREM metallicity and continuum subtraction
368:  \section{Oxygen Abundance}
369:        In order to estimate the abundance of oxygen
370:        we employ three  methods using relations of various
371:        emission-line fluxes: the R$_{23}$ strong-line abundance
372:        calibration  of \citet{mcgaugh},
373:        the O3N2 strong-line calibration of
374:        \citet{pp04} and the Bayesian metallicity estimates of \citet{trem05}.
375: 
376:         We use the following flux ratios in our calculations dereddened using the extinction curve of Seaton (1979), assuming R$_v=3.1$ and 
377:         Case B recombination \citep{oster}:
378:              \begin{equation}
379:       \frac{[OIII]}{[OII]}\equiv\frac{[OIII]\lambda5007}{[OII]\lambda3727}
380:               \end{equation}
381:               \begin{equation}
382:         O_{32}\equiv\frac{[OIII]\lambda5007
383:                  +[OIII]\lambda4959}{[OII]\lambda3727}
384:               \end{equation}
385:                \begin{equation}
386: \frac{[OIII]}{H\beta}\equiv\frac{[OIII]\lambda5007}{H\beta}
387:               \end{equation}
388:               \begin{equation}
389: \frac{[NII]}{H\alpha}\equiv\frac{[NII]\lambda6584}{H\alpha}
390:               \end{equation}
391:               \begin{equation}
392: \frac{[NII]}{[OII]}\equiv\frac{[NII]\lambda6584}{[OII]\lambda3727}
393:               \end{equation}
394:              \begin{equation}
395: R_{23}\equiv\frac{[OII]\lambda3727+[OIII]\lambda5007+[OIII]\lambda4959}{H\beta}
396:               \end{equation}
397:              \begin{equation}
398:  O3N2\equiv\log( \frac{[OIII]\lambda5007/H\beta)}{[NII]\lambda6583/H\alpha})
399:               \end{equation}
400:                
401:         The strong line abundance calibration was first
402:        developed by \citet{pag79} and \citet{alloin}. 
403:        The various line ratios that have been used to
404:        calculate abundances are [NII]$/$H$\alpha$,
405:        [OIII]$/$[NII],
406:        [NII]$/$[OII], ([SII]
407:        $\lambda\lambda$66717,6731+ [SIII]
408:        $\lambda\lambda$9069,9532])$/$H$\beta$, and R$_{23}$,
409:         which was first introduced by \citet{pag79}. Oxygen strong line
410:        abundance calibrations are either (1) based on
411:        photo-ionization models \citep{diaz,kd03} or (2) on abundances
412:        measured in nearby HII regions where the electron
413:        temperatures of the ionized regions can be measured. The latter method
414:        requires detection of faint auroral
415:        emission lines (e.g.,[OIII]$\lambda$4363, [NII]$\lambda$5755) to determine the electron
416:        temperature, {\it T$_e$}. The empirical strong-line analytical
417:        expressions are created from these measurements to
418:        estimate the abundance in galaxies and HII regions
419:        that lack significant detections of the auroral lines
420:        but have similar abundances.
421: 
422:         Recently the studies of
423:        \citet{kbg,bgk,gkb} using electron temperatures for
424:        high metallicity HII regions in M101 and M51,
425:        indicate that various strong-line methods calibrated to photo-ionization models (e.g,
426:        \citet{mcgaugh}) estimate a higher abundance at high
427:        metallicities by 0.2-0.5 dex than the {\it $T_e$} method abundances.
428:        It is currently not clear which method is correct. There is some evidence that the abundances calculated by the {\it $T_e$} method 
429:      may be underestimated due to temperature fluctuations in the ionized regions causing the electron temperatures to be overestimated,
430:      and that the strong line abundances may be more correct since their line ratios are not as temperature sensitive as 
431:      [OIII]4363 \citep{peimbert, bres, bgk,kbg}. 
432:   
433:       In our analysis we use two strong-line calibrations to estimate the oxygen abundance: the R$_{23}$ diagnostic of \citet{mcgaugh} calibrated 
434:       from photo-ionization models,
435:       and the strong-line ([OIII]/H$\beta$)/([NII]/H$\alpha$) diagnostic of \citet{pp04} calibrated from {\it $T_e$} abundances. 
436:         The derived abundances for each method are respectively labeled M91, and O3N2 throughout the 
437:       remainder of this work. We add to these strong line abundances the strong line oxygen abundance estimates of 
438:        \citet{trem05} (hereafter labeled T04), obtained from
439:        likelihood distributions of oxygen abundances derived by
440:        matching emission line fluxes from integrated galaxy spectra
441:        models of \citet{cl01} to the measured fluxes.     
442: 
443: 
444:       The R$_{23}$ analytical
445:       expressions calibrated by \citet{mcgaugh} and given in
446:       \citet{kobul} that we use to calculate M91 are cited below. Many
447:       other authors have developed techniques for estimating
448:       abundances from R$_{23}$. Examples include \citet{cl01},
449:       \citet{alloin}, and \citet{EnP}. R$_{23}$ is useful
450:       because it provides an estimate of the total cooling due
451:       to oxygen. The major caveat  with $R_{23}$ is that it is
452:       double valued with respect to metallicity. At low oxygen abundances, $12+\log O/H\lesssim8.4$, R$_{23}$
453:       increases with rising abundance until $12+\log O/H\gtrsim8.4$, after which it begins to
454:       decrease as metals begin to cause efficient cooling,
455:       lowering the electron temperature and thus decreasing
456:       the amount of collisional excitation of the oxygen ions. The metal-poor branch expression is:
457:       
458:      \begin{equation} 
459:          12 + \log(O/H) =7.065+.767x+.602x^2- y(.29
460:          +.332x-.3318x^2)
461:              \end{equation}
462:     and the metal-rich branch expression is:
463:         \begin{equation}
464:          12+\log(O/H) = 9.061-.2x-.237x^{2}-.305x^{3}-.0283x^{4}- 
465:               y(.0047-.0221x -.102x^{2} -.0817x^{3}-.00717x^{4})
466:          \end{equation}
467:  
468:         where x$\equiv\log(R_{23}$) and
469:         y$\equiv\log(O_{32}$). O$_{32}$ is used to correct
470:         the effect of the ionization parameter  on R$_{23}$. \citet{kd03}
471:         have found that the O$_{32}$ ratio depends on
472:         metallicity and as a result is not a good indicator of ionization
473:         unless an initial estimate of metallicity can be
474:         given and an iterative process is applied.   
475: 
476:        To determine on which branch the correct solution
477:        lies, we use the metallicity sensitive ratios [NII]$/$H$\alpha$ and
478:        [NII]$/$[OII]. For
479:        $\log([NII]/H\alpha$) $<$-1 and $log([NII]/$[OII]) $<$ -1.5
480:        we use the metal-poor expression. For $\log[NII]/H\alpha$ $>$-1 and $\log[NII]/$[OII] $>$ -0.8   
481:        we use the metal-rich expression. If  -0.8$>\log(
482:        [NII]/$[OII]) $>$ -1.5, then  we use the  $\log([NII]/H\alpha)$ ratio as stated above to determine the correct branch. 
483:         Where the two ratios give conflicting
484:        estimates, the average of the two expressions is used to
485:        derive the abundance. This is because the solutions for the two branches converge at intermediate metallicities, 
486:       $12+\log O/H\sim8.4$, and
487:        it is near this metallicity where the
488:        metallicity sensitive ratios  are likely to  give a conflicting answer. The average of the two branch solutions in this 
489:        case should minimize any bias
490:        in the calculations.
491:        A caveat with this procedure is that galaxies with intermediate oxygen abundances 
492:        but with high SFRs will have lower ratios of [NII]/H$\alpha$ and as a result
493:        can have their abundances calculated with the lower branch and therefore underestimated.
494:        
495:        
496:       The strong line calibration O3N2  developed by
497:       \citet{pp04} is shown below. The calibration based on
498:       this flux ratio also has several problems. First, it is not corrected for ionization parameter. Second, it is
499:       based on the flux from a forbidden nitrogen line whose
500:       abundance many authors claim depends star-formation history of
501:       the galaxy. As a result the calibration is accurate to
502:       $\log (O/H) =\pm0.25$, and is only valid for
503:       O3N2$< 1.9$ (e.g. $12 + \log(O/H) \gtrsim 8.1$). 
504:        \begin{equation}
505:   12 +\log(O/H) = 8.73-0.32 \times O3N2      
506:        \end{equation}
507:        There is also evidence that at metallicities $\gtrsim
508:        \log(O/H)_{\odot}$ (O3N2 $\lesssim 0.4)$ the O3N2
509:        calibration overestimates the
510:        oxygen abundance \citep{bgk}.            
511:    
512:          %%%%%%%%%%%REEEEEDDDDOOO PARAGRAPH
513:        Figure 1 shows the difference between all  
514:        the abundance calibrations as a function of stellar mass. 
515:        In the figure we have transformed each panel into a 75 by 75 pixel image. The mean SFR/M$_*$ of the points in each pixel 
516:        is shown in true color representation. The mean difference between M91 and O3N2 shows some dependence on stellar mass, with the lower 
517:           branch of M91 giving lower abundances than O3N2, typically
518:         about $0.1$ dex with a dispersion of $0.18$ dex. The upper branch of M91 calculates larger abundances than O3N2 typically by $\sim0.2$ dex with a
519:         dispersion of $0.14$ dex. 
520:        While the offset between O3N2 and M91 shows little dependence on stellar mass, 
521:       the offset between T04, and the M91, O3N2 diagnostics show a dependence on stellar mass. As galaxy mass increases, 
522:       T04 estimates an increasingly larger metallicity than the other two calibrations.
523:         
524: \section{Nitrogen Abundance}
525:     We calculate nitrogen abundance estimates by first using the calibration of \citet{thurston} to estimate the temperature in the
526:     [NII] emission region using their calibrated empirical relation created from photo-ionization models:
527:          \begin{equation}
528:                    t_{[NII]}=0.6065+0.1600x
529:                    +0.1878x^2+0.2803x^3
530:             \end{equation}
531:      
532:     where $x\equiv\log R_{23}$.
533:     We then use this
534:     temperature to determine the the ratio of N$^+$/O$^+$ based
535:     on the empirical calibration of \citet{pag92} based off of {\it T$_e$} abundances:
536:                
537:                   \begin{equation}
538:                      Log\frac{N^+}{O^+} = Log
539:                      \frac{[NII]}{[OII]} + .307 - .02Log t_{[NII]} - \frac{0.726}{t_{[NII]}}    
540:                   \end{equation}
541:       
542:       We finally assume that N$/$O=N$^+/$O$^+$. \citet{thurston} found
543:       through modeling that this assumption is reliable, with only small uncertainties,
544:       $\sim$.05 dex. \citet{garnet90} concurs that the  N$^+/$O$^+$
545:       is an accurate N$/$O indicator for low abundances or
546:       where the ionizing stars are hotter than
547:       40,000K. Results of modeling by \citet{stasinka} show that even at high abundance, equating the ion
548:       ratio to the element ratio is good to within 5\%. 
549:  
550:        
551:        We also calculate the nitrogen and oxygen  abundance via the {\it T$_e$} method
552:           for the 33 objects in our {\sl GALEX} emission line sample having at least a
553:        3$\sigma$ detection of [OIII]43643 to ensure reliable estimates of the electron temperature in the [OIII] ionization regions. Table 2 
554:        shows the derived ${\it T_e}$ abundances and derived galaxy parameters from SED fitting for these objects. We use the TEMDEN
555:       procedure in the IRAF package Nebular \citep{fivel} to derive the electron temperature from the ratio of ([OIII]5007+[OIII]4959)/[OIII]4363. 
556:       The electron temperature in the [OII] and [NII] regions were then estimated using the linear relation from Garnett (1992) to convert the mean 
557:      [OIII] electron temperatures into mean  electron temperatures in the [OII] ionization regions. We then assume that since [NII] and [OII] have relatively
558:      similar ionization energies that the [NII] electron temperature equals the [OII] electron temperature. The abundance of each ion O$^{2+}$, O${^+}$, 
559:      and N$^+$ were then calculated using the IONIC procedure in Nebular. 
560:      
561:       All four oxygen calculations show a small abundance range for this sub-sample of objects, thus limiting our ability to determine
562:       if the difference between the two methods has any dependence on abundance or on N/O. 
563:       The N/O ratio calculated with the strong line calibration
564:       shows that it is typically $\sim0.1$ dex greater than the ratio determined by the {\it $T_e$} method with a dispersion of $0.07$ dex.  
565:       How accurate the strong line N/O ratio is for higher oxygen abundances is unknown, and its precision is
566:       lacking. The mean error on log N/O for the entire sample is 0.17 dex, due mostly  to the error on the R$_{23}$ temperature. 
567:       For the the purposes of the rest of our analysis, the accuracy of the strong line nitrogen diagnostic does not matter,
568:       only the relative difference between each galaxy is of importance.   
569:    
570: 
571: \section{The N/O versus O/H Relationship}                     
572:             The time delay scenario for the production of nitrogen predicts that starbursting galaxies exhibit a rise in oxygen abundance
573:           along with a drop in N/O \citep{contini, vzee06, henry99}. The addition of UV data from {\sl GALEX} 
574:           to the 5 band SDSS photometry makes it possible to distinguish between galaxies recently hosting
575:           starbursts and those with declining star formation, because the FUV passband is responsive to star formation
576:           on timescales of 10 Myr and the NUV passband on timescales of 100 Myr \citep{martin, bc03}. We use our O/H and N/O estimates along with the 
577:           results from the Bayesian broad band SED analysis 
578:           to examine if the relative abundance of nitrogen to oxygen in a galaxy can be explained by the galaxy's star formation history. 
579:                       
580:            The relationship between N/O and O/H for our sample is shown
581:           in Figure 2. The points have been pixelated and then scaled by color
582:           to show the mean value of specific star formation rate
583:           of the points in each pixel. The specific star formation rate indicates the relative number of recently formed ($\sim100Myr$) high mass stars
584:           to the cumulative number of stars formed over a galaxy's star formation history. 
585:           Galaxies with large specific star formation rates have recently undergone a burst of star formation 
586:           or have a slowly declining SFR. The mean standard deviation in  SFR/$M_*$ for each pixel is $\sim0.07$ dex.
587:            In Figure 3 we plot our nitrogen-oxygen relationship again
588:           with the data points plotted as a shaded 2D histogram to aid the interpretation of the previous figure.
589: 
590:            In the figures we  have included the simple closed box 
591:           model of \citet{villa93} for the primary (solid
592:           line), secondary (dashed line), and primary +
593:           secondary (dashed-dotted line) production of
594:           nitrogen. This model assumes that nitrogen has both a
595:           primary and secondary component, and that oxygen has
596:           only a primary component.  The time rate of
597:           change of each element is taken to be proportional
598:           to the star formation rate which is assumed to equal a constant times the fraction
599:           of galaxy's mass in gas ($=1$ at t$=0$).
600:            Assuming that there are 
601:           no time delays in the release of the material, a solution for the model can be 
602:          found,  $\log [N/O] = \log[a+b \times [O/H]]$, 
603:           where a is the primary yield of nitrogen divided by
604:           the yield of oxygen and b is the secondary yield of
605:           nitrogen divided by the yield of oxygen. \citet{villa93} quote
606:           values of a=.034 and b=120 using a  by-eye fit
607:           to line strengths taken from literature  for HII
608:           regions in nearby galaxies. 
609:           
610:           The three oxygen abundance methods, allowing for the relative offsets between each method,
611:            are all consistent with galaxies containing primary nitrogen 
612:            at low metallicities and a secondary component at higher metallicities. The mean scatter of N/O as a
613:            function of oxygen abundance is 0.08 dex for O3N2, 0.11 dex for M91 and 0.13 dex for T04.
614:           We note that these three derivations for oxygen abundance are not completely independent of the nitrogen abundance.
615:           The O3N2 value depends on a flux ratio containing [NII]6584, while $R_{23}$ is used to calculate 
616:           the [NII] temperature required to determine the N/O ratio. Furthermore, 
617:           the models used to derive T04 have prior distributions of metallicities where nitrogen abundance is selected to have a only a primary dependence on
618:           the oxygen abundance below $12+\log$ O/H$<8.25$, and a completely secondary dependence for metallicities greater than this.
619:            T04 and O3N2 diagnostics are therefore predetermined to exhibit secondary nitrogen production, 
620:            and are not useful in determining
621:           the relative amount of secondary or primary nitrogen in each galaxy. Furthermore, the abundances determined by T04  may
622:           slightly overestimate abundances for galaxies that have an increased N/O ratio from primary+secondary nitrogen. The likelihood estimates of T04
623:           depend on the flux from nitrogen emission lines,  but only consider that nitrogen is secondary in origin, and do not accurately account for
624:           nitrogen fluxes from galaxies containing secondary plus primary nitrogen. Of the three diagnostics, the M91 calculation 
625:            has the least dependence 
626:            on the N/O diagnostic. This is because the N/O calculation depends slightly on the temperature estimate obtained by $R_{23}$, which introduces
627:          a scatter in N/O that increases from $\sim0.04$ dex at the lowest values of N/O to $\sim0.1$ dex at highest N/O values. The
628:           fact that O3N2 and T04 show a similar secondary dependence on N/O as M91 is an indication that the interdependence between the N/O 
629:           calculation and T04 or O3N2 is only a small effect.
630: 
631: 
632: %%SUB HEADING BREAK
633: \subsection{N/O and SFR/M$_*$}
634: 
635:          The main results emerge when we consider the relationship between N/O and O/H as a function of specific star formation rate.
636:           Figure 2 gives several interesting results. First, for galaxies with high abundances ($12+\log$ O/H$\gtrsim8.6$),
637:           the M91 and T04  diagnostics both indicate that for galaxies with similar O/H, the most extreme starbursts (highest values of SFR/$M_*$) tend to have
638:           lower N/O. This is shown more clearly in Figures 3 and 4. In these figures we divide our emission sample into sub-samples of 
639:           specific star formation rate:   
640:            log SFR/M$_* < -10.1$ (red points),  $-10.1<$log SFR/M$_* <-9.1$ (green points), and log SFR/M$_* > -9.1$ (blue points). We then calculate the mean 
641:          N/O value and error on the mean for each of the sub-samples in increments of $0.1$ dex in $\log$ O/H. 
642:          Figure 3 plots the mean N/O values for each sub-sample 
643:          as a function  of metallicity, and figure 4 plots the difference between the mean N/O value in each sub-sample  
644:           and the mean N/O value for the entire sample. Tables 3, 4 and 5 
645:          list the mean values of N/O, the errors on the mean,
646:          and the number of galaxies for each bin of O/H with more than 30 galaxies in each sub-sample. All three diagnostics show that the galaxies
647:          with  log SFR/M$_* > -10.1$ have lower N/O values than galaxies in the lowest specific star formation rate sub-sample (log SFR/M$_* < -10.1$)
648:          in each decrement of metallicity between $8.5$ and $9.0$ dex. As metallicity
649:          increases, and nitrogen becomes largely secondary in origin, and 
650:          the difference between the N/O ratios of the sub-sample with the lowest specific star formation rates and the other two sub-samples  decreases.
651:                     
652:           The M91 and T04 diagnostics also show that the most extreme starbursts (log SFR/M$_* > -9.1$) 
653:           at intermediate metallicities on average have N/O ratios $0.02$ dex lower than galaxies with average specific 
654:          star formation rates, $-10.1<$log SFR/M$_* <-9.1$, a decrease in N/O of $\sim3\%$.
655:           The O3N2 diagnostic shows the opposite trend of the other two diagnostics since O3N2
656:           is not corrected for ionization parameter. 
657:           At the lowest and highest metallicities
658:           no difference is found between the N/O ratios of the two sub-samples in all three diagnostics.
659:  
660: %%%%%%
661: %with high  log SFR/M$_*$ have lower N/O ratios because the chemical enrichment of the galaxies ISM is dominated
662: % by high mass stars that release more oxygen than nitrogen. The chemical enrichment of  galaxies with low log SFR/M$_*$, which are not currently forming as many 
663: % high mass stars, are dominated by the products of intermediate mass stars, which create more nitrogen than oxygen.
664: 
665:             Our findings are consistent with  similar conclusions reached by \citet{contini}.            
666:          The galaxies with the lower specific star formation rates
667:            have the highest N/O ratios because they  are currently forming comparatively fewer
668:           high mass stars. This allows the chemical enrichment of the galaxy to be dominated by the products of intermediate mass stars, 
669:           which generate  more nitrogen than oxygen, causing N/O in these galaxies to rise.
670:            At low metallicities no difference is found between 
671:            the N/O ratios of the  the most extreme starbursts and the average star-forming sub-samples. This is because 
672:            there are only a small number of galaxies in our sample with low metallicities
673:            and most of these are in the sub-sample with the highest specific star formation rates. At high metallicities, no difference is found  between the 
674:            the most extreme starbursts and the average star-forming sub-sample. This is presumably because the oxygen generated by  
675:            the high mass stars formed during the latest starburst constitutes only a small fraction of the total oxygen abundance of the galaxy 
676:            and has little effect on lowering the N/O ratio. At intermediate metallicities the oxygen abundance is relatively small,
677:            such that the oxygen created in a starburst
678:            constitutes  a large fraction of the oxygen abundance of a galaxy and causes a larger decrease in the N/O ratio.  
679:        
680:          Other possible explanations for the N/O ratios are variable IMFs and galactic winds. 
681:          An IMF that produces more massive stars for  galaxies  with higher  specific star formation rates, could possibly cause the low N/O ratios
682:        of the strongest  starbursts  seen in Figures 2,3, and 4. 
683:           Such a variable IMF could plausibly have a slope parametrized by either metallicity, SFR, or both. 
684:           \citet{silk} conjectures that the IMFs of starbursts may be weighted to
685:           form more massive stars, and several authors have previously parametrized IMFs with a dependence  on metallicity \citep{matt_t, scully}. 
686:           At this time, the validity of a variable IMF and its effect on the abundances cannot be assessed.
687:            We find no need to invoke a variable  IMF to model the UV and optical SEDs  of these galaxies. Furthermore,
688:            \citet{chia} have found that that chemical 
689:          evolution models for the Galaxy that use a metallicity dependent IMF
690:           do not adequately reproduce the observational constraints of the solar neighborhood. 
691:             
692:                   
693:          Galaxies with high specific star formation rates could also have galactic winds that differentially remove one element with respect to the other.
694:          With regards to differential flow of oxygen, 
695:          \citet{vzee06} examined the ratio of N/O for a sample of dwarf galaxies, and argues that their data suggests that either differential outflow 
696:          of oxygen occurs in every  galaxy in their  sample
697:          with the same efficiency or that differential outflow of oxygen has a negligible effect on N/O ratios.  
698:          They found that the correlation of oxygen abundance with optical luminosity for their sample had a lower 
699:          scatter than the correlation of nitrogen abundance with optical luminosity. They argue that if differential outflow was the cause of the scatter 
700:          in N/O, then the oxygen-luminosity correlation should have a larger scatter than the nitrogen-luminosity correlation, since the
701:          the outflow of oxygen would depend on other galaxy parameters such as galaxy mass, and ISM structure. Nitrogen may be differentially removed in 
702:          galaxies that have high specific star formation rates, but there is no reasonable explanation as to why this might occur. 
703:          In fact, one would expect the opposite, that  
704:          oxygen and not nitrogen  would be differentially removed in starbursts since the kinetic energy responsible for ejecting the material likely comes from
705:           the supernovae of high mass stars that produce very little nitrogen with respect to oxygen.   
706:         
707: \subsection{N/O vs M$_*$ and {\it g-r} optical color}           
708:            In order to further test the above explanation, we plot the nitrogen to oxygen  relationship again,
709:            with the pixels scaled by color with mean values of stellar mass, and {\it g-r} color in Figures 5 and 6 respectively.  \citet{trem05}
710:           found a tight correlation of $0.1$ dex between increasing stellar mass and oxygen abundance, so we expect that mass will increase
711:            with O/H. Figure 5 shows that the stellar mass increases with O/H with little dependence on N/O for all three methods. 
712:           Along the same lines, \citet{vzee06} analyze a dwarf galaxy sample and conclude that a trend of increasing N/O 
713:              correlates with redder B-V color and hence, lower star formation rate.  
714:            Based on this result, we expect that increasing values of N/O should correlate with redder g-r color.  
715:             In figure 6 we see that as N/O increases 
716:           the average {\it g-r} color increases for T04, M91, and O3N2 at high metallicities. At metallicities below $12+\log$ O/H$\sim8.4$ both 
717:           M91, and T04 show this
718:           trend. O3N2 does not because it is not corrected for ionization effects. All methods of determining the oxygen abundance show a dependence
719:           on mass irrespective of their nitrogen abundance. The T04 and M91 diagnostics 
720:           show that galaxies of similar metallicity but with higher N/O values
721:           have redder {\it g-r} color. These results suggest that the trend between higher specific star formation rates and lower N/O values is a real trend, 
722:           but more reliable and consistent metallicity diagnostics are required to test this result.  
723: 
724: 
725: \subsection{{ \it T$_e$} sample  N/O vs O/H}          
726:          As a further check, we use the ${\it T_e}$ method abundances to determine whether the trend 
727:          for galaxies with higher SFR/M$_*$ to have lower N/O ratios is genuine, and not 
728:          produced by the strong line abundance calculations themselves due to a dependence on an unknown  galaxy parameter.
729:          The abundances derived from the {\it $T_e$} method are dependent only on electron
730:          temperature and density, and the N/O ratio calculated by this method is not predetermined to show secondary dependence.   
731:           In Figure 7 we show the N/O ratio versus  O/H for the 33 galaxies with abundances measured
732:           by the {\it $T_e$} method. In the upper left the points are colored by their {\it g-r} optical colors.
733:           Even though there is a good deal of scatter in the figure, the galaxies that
734:           have the lowest N/O ratios tend to be the bluest in {\it g-r}, with the mean value of ${\it g-r} = 0.12$ and 
735:           a standard error on the mean of $.04$ for galaxies with 
736:          $\log N/O < -1.5$ and  {\it g-r} $= 0.2$ with a standard error on the mean of $0.04$ for galaxies with $\log N/O >-1.5$.  These galaxies also tend to have
737:           slightly higher specific star formation rates  with a mean difference of $0.12$ dex between galaxies with  
738:          $\log N/O < -1.5$  and galaxies with  
739:          $\log N/O > -1.5$.  The galaxies with lower N/O ratios also have slightly higher H$\alpha$ equivalent widths, which is an indicator of the current 
740:           star formation relative to past star formation, on timescales of 10 Myr. 
741:            The SED fitting indicates that to a 95\% reliability at least half 
742:           of these galaxies formed 1\% (and as much as $\sim50\%$) of their stellar mass in bursts within the last 100Myr.
743:            If there is indeed a time delay between the release of oxygen from the massive stars 
744:           and nitrogen from the intermediate mass stars, then these starburst galaxies should have an influx of newly synthesized oxygen that 
745:           will raise the oxygen abundance and reduce the N/O ratio.  The results from the {\it $T_e$} abundances
746:             slightly favor this scenario, but due to the small 
747:           sample of galaxies with 3$\sigma$ detections of [OIII]4363, the small range of specific star formation rates,
748:            and the uncertainties on the abundances, we are unable to discern if the star formation history is really the cause 
749:           of the scatter of N/O values  for the {\it $T_e$} sample. 
750:          We would expect that since all of the 33 galaxies are large starbursts, that the N/O ratios would lie close to the secondary
751:           nitrogen curve. The explanation posited by \citet{iz05} for the scatter in N/O for galaxies with similar metallicities 
752:           is that the Wolf Rayet stars in these galaxies have released a significant amount of nitrogen from winds, which 
753:           being an order of magnitude more dense than the surrounding ISM  can cause the N/O ratio to appear high, increasing by as much as $0.23$ dex.
754:            The N/O ratio will decrease as the nitrogen from the WR winds has time to diffuse
755:           into density equilibrium with the ISM, raising the overall ISM log N/O ratio by $0.03$ dex. However, the chemical evolution models of \citet{chia03} show 
756:           that the scatter in N/O ratios of dwarf galaxies at metallicities similar to those in our  
757:            our {\it $T_e$} sample can be explained by different star formation histories, 
758:            different burst strengths, and burst durations; they show that there is no need to invoke nitrogen from winds of massive stars to show this effect.
759:           The galaxies with $\log N/O >-1.5$ have on average
760:            only slightly  bluer optical colors than the rest of the {\it $T_e$} galaxies. This indicates that intermediate mass
761:          stars from the last major star formation event may be responsible for the high N/O ratios, but does not rule out that winds could cause 
762:          a portion of the observed scatter, (though the errors on the measured abundances 
763:           are able to account for a large portion of the observed scatter in N/O). To resolve this issue, and effect of variable IMFs or 
764:           other dynamical processes on N/O such as
765:           mixing timescales of the newly synthesized material, 
766:           more reliable and consistent metallicity measurements             
767:           are needed,  with errors in the derived abundance less than
768:           $0.1$ dex. New nitrogen diagnostics 
769:           for metal rich galaxies are particularly needed to compare our results obtained using the \citet{pag92} strong line diagnostic.
770:          These results should also be compared to chemical evolution models to substantiate their validity.
771: 
772: 
773: \section{Conclusion}
774:    We consider galaxies detected by
775:    {\sl GALEX} in the Medium Imaging Survey to a limiting magnitude of $NUV=23 (AB)$. We match our UV star-forming galaxies
776:    to ${\it z}=0.3$ with the SDSS DR4 spectroscopic sample. \\ \\
777: 
778:     1. {\sl GALEX} at MIS depth ($NUV_{limit}\sim23.0 AB$) 
779:        detects 98.4\% of star forming SDSS galaxies in the DR4 spectroscopic sample matching our emission line criteria of
780:        5$\sigma$ detections of H$\alpha$,
781:        H$\beta$, [NII]6584, and 3$\sigma$ detections of [OIII]5007 and [OII]3727. \\  \\ 
782:  
783:      2. For our emission line sample of $\sim8,000$ {\sl GALEX}/SDSS star forming galaxies, 
784:     we have examined each galaxy's oxygen abundance for three strong line abundance measurements. These are calibrated off photo-ionization models, M91, 
785:    {\it $T_e$} determined abundances, O3N2,
786:    and the Bayesian likelihood estimates, T04. We compare each abundance method as a function of both M$_*$ and SFR/M$_*$. 
787:     Compared to the other two methods O3N2 is found to increasingly estimate lower oxygen abundances 
788:     for galaxies with higher SFR/M$_*$ since it is not corrected for ionization parameter.\\ \\   
789: 
790:     3. We investigate the relationship between N/O and O/H 
791:     using the three different O/H diagnostics and the strong line calculation of N/O from \citet{pag92}. We use the specific SFR  
792:     derived from SED fits to the $7-$band {\sl GALEX}$+$SDSS photometry to indicate of the strength of the starbursts in each galaxy
793:     over the last 100Myr.\\ \\  
794: 
795:     4. Star forming galaxies that are currently forming a
796:        large percentage of their stellar mass, as parametrized by SFR/M$_*$, have smaller  values of N/O at a given metallicity for 
797:        for all three metallicity diagnostics, supporting the results of \citet{contini}.  
798:        This trend spans the metallicity range of $\sim0.6$ dex from $12+log O/H\sim8.4$ to  $12+log O/H \sim9.0$~dex and suggests the scenario
799:         that the scatter in N/O ratio for galaxies of similar metallicities is due to the ratio of current to past averaged SFR.
800:         The observed effect is modest, since the change in N/O is of the order of the abundance uncertainties. However the abundance dispersion could also arise
801:     due to varying mixing times of the newly synthesized oxygen into the ISM.   More realistic  and consistent metallicity
802:         diagnostics are required to further test this result. \\
803:      
804: 
805: \acknowledgments
806: 
807: GALEX is a NASA Small Explorer, launched in April 2003. We
808: gratefully acknowledge NASA's support for construction,
809: operation, and science analysis for the GALEX mission,
810: developed in cooperation with the CNES of France and the
811: Korean Ministry of Science and Technology. Funding for the
812: creation and distribution of the SDSS Archive has been
813: provided by the Alfred P. Sloan Foundation, the Participating
814: Institutions, NASA, NSF, DoE, Monbukagakusho, and the Max
815: Planck Society.
816: 
817: 
818: {\it Facilities:} \facility{GALEX}
819: 
820: 
821: \begin{thebibliography}{}
822: %\bibitem[Abazajian et al.~(2004)]{abaz} Abazajian, K. et
823: %  al. 2004, \aj, 128, 502
824: \bibitem[Alloin et al.~(1979)]{alloin} Alloin, D.,
825:   Collin-Souffrin, S., Joly, M., \& Vigroux, L. 1979, A\&A,
826:   78, 200
827: %\bibitem[Alongi et al. (1993)]{alongi} Alongi M., Bertelli F.,
828: %  Bresssan A., Chiosi C., Fagotto F., Greggio L., Nasi E.,
829: %  1003, A\&ASS, 97, 851
830: \bibitem[Baldwin,  Phillips, \& Terlevich
831:   (1981)]{bpt} Baldwin, J. A.,  Phillips, M. M., \&
832:   Terlevich, R. 1981, PASP, 93, 5
833: %\bibitem[Bell, E. F. et al. (2004)]{bell04} Bell, E. F. et
834: %  al. 2004, ApJ, 608, 752
835: \bibitem[Bertin, \& Arnouts (1996)]{bar} Bertin, E.,
836:   \& Arnouts, S. 1996, A\&AS, 117, 393 
837: %\bibitem[Blanton et al (2003)]{blanton} Blanton, M. et al
838: %  2003, \aj, 125, 2276
839: %\bibitem[Bressan et al. (1993)]{bressan} Bressan A., Fagotto
840: %  F., Bertelli G., Chiosi C., 1993, A\&AS, 100, 647
841: 
842: \bibitem[Bresolin (2006)]{bres} Bresolin, F. 2006, astro-ph/0608410
843: \bibitem[Bresolin, Garnett, \& Kennicutt (2004)]{bgk} Bresolin,
844:   F., Garnett, D. R. \& Kennicutt, R. C. 2004, ApJ, 615, 228 
845: %\bibitem[Brinchmann et al. (2004)]{jarle} Brinchmann, J.,
846: %  Charlot, S., White, S. D. M., Tremonti, C., Kauffmann, G.,
847: %  Heckman, T., \& Brinkmann, J., MNRAS, 351, 1151
848: %\bibitem[Bower, Lucey, \& Ellis,
849: %   (1992)]{Bower} Bower, R. G., Lucey, J. R., \& Ellis,
850: %  R. S. 1992, MNRAS, 254, 601
851: \bibitem[Bruzual \& Charlot (2003)]{bc03} Bruzual,
852:   G., \& Charlot, S. 2003, MNRAS, 344, 1000 
853: \bibitem[Chiappini, Matteucci \& Padoan
854:   (2000)]{chia}Chiappini, C., Matteucci, F., \& Padoan,
855:   P. 2000, ApJ, 528, 711
856: \bibitem[Chiappini, Romano, \& Matteucci (2003)]{chia03} Chiappini, C., Romano, D., \& Matteucci, F. 2003, MNRAS, 339, 63 
857: \bibitem[Chaippini, Matteuci, \& Ballero (2005)]{chia05}Chiappini, C., Matteucci, F., Ballero, S. K. 2005, A\&A, 437, 429
858: \bibitem[Chaippini et al. (2005)]{chia06}Chiappini, C., Hirschi, R., Meynet, G.,  Maeder, A., \& Matteucci, F. 2006, A\&A, 449, 27
859: \bibitem[Charlot \& Longhetti (2000)]{cl01} Charlot,
860:   S., \& Longhetti 2000, MNRAS, 323 887
861: \bibitem[Charlot \& Fall (2000)]{cf00}  Charlot,
862:   S., \& Fall, S. M. 2000, ApJ 539, 718
863: \bibitem[Consid$\tilde{e}$re et al.~(2000)]{considere} Considere, S., Coziol, R., Contini, T., \& Davoust,
864:   E. 2000, A\&A, 356, 89
865: \bibitem[Contini et al.~(2002)]{contini} Contini, T.,
866:   Treyer, M. A., Sullivan, M., \& Ellis, R. S. 2002, MNRAS,
867:   330, 75
868: %\bibitem[Dekel \& Woo (2003)]{dek} Dekel, A. \& Woo, J. 2003,
869: %  MNRAS, 344, 1131
870: %\bibitem[De Robertis et al. (1987)]{derobertis} De Robertis,
871: %  M. M., Dufour, R. J., \& Hunt, R. W. 1987, JRASC, 81, 195
872: \bibitem[Diaz et al.~(2000)]{diaz} Diaz, A. I., Castellanos,
873:   M., Terlevich, E., \& Garcia-Vargas, M. L. 2000, MNRAS, 318, 462 
874: \bibitem[Diaz \& Tosi (1986)]{diaz86} Diaz, A. I., \& Tosi, M. 1996, A\&A, 158, 60
875: \bibitem[Edmunds \& Pagel (1984)]{EnP}
876:   Edmunds, M. G., \& Pagel, B. E. J. 1984, MNRAS, 211, 507
877: %\bibitem[Fagotto et al. (1994a)]{faga} Fagotto F.,  Bressan
878: %  A., Bertelli G., Chiosi C., 1994a, A\&AS, 104, 365
879: %\bibitem[Fagotto et al. (1994b)]{fagb} Fagotto F.,  Bressan
880: %  A., Bertelli G., Chiosi C., 1994a, A\&AS, 105, 29
881: \bibitem[Fields \& Olive (1998)]{field} Feilds, B. D., \& Olive, K. A. 1998, ApJ, 516, 797
882: \bibitem[Fukugita et al.~(1996)]{fuk96} Fukugita, M.,
883:   Ichikawa, T., Gunn, J. E., Doi, M., Shimasaku, K.,
884:   \& Schneider, D. P. 1996, \aj, 111, 1748
885: %\bibitem[Garnett (2004)]{garnetbook} Garnet, D. R. 2004,
886: %  in Evolution and Origin of the Elements, Ed.  McWilliam, A. ,
887: %  \& Rauch, M. (Cambridge University Press),  405
888: %\bibitem[Garnett et al (2004)]{garnet04} Garnet, D. R.,
889: %  Edmunds, M. G., Henry, R. B.C., Pagel, B. E., Skillman,
890: %  E. D. 2004, ApJ, 128, 2772
891: \bibitem[Garnett, Kennicutt \& Bresolin (2004)]{gkb} Garnett,
892:   D. R., Kennicutt, R. C., \& Bresolin, F. 2004, ApJ, 607, 21
893: %\bibitem[Garnett (2002)]{garnet02} Garnett, D. R. 2002,
894: % Apj, 581, 1019
895: %\bibitem[Garnett (1992)]{garnet92} Garnett, D. R. 1992, AJ,
896: %  103, 1330
897: \bibitem[Garnett (1990)]{garnet90} Garnett, D. R. 1990,
898:   \aj, 363, 142
899: %\bibitem[Girardi et al. (1996)]{girardi} Girardi L., Bressan
900: %  A.,  Chiosi C., Bertelli G., Nasi E., 1996, A\&AS, 117, 113
901: %\bibitem[Heckman et al. (2000)]{heck} Heckman,
902: %  T. M.,  Lehnert, M. D.,  Strickland, D. K., \& Armus,
903: %  L. 2000, ApJS,  129, 493
904: \bibitem[Henry \& Worthey (1999)]{henry99} Henry, R. B. C.,  \&
905: Worthey, G. 1999, PASP, 111, 919
906: \bibitem[Hogg et al.~(2001)]{hogg} Hogg, D. W.,
907:   Finkbeiner, D. P., Schlegel, D. J., \& Gunn, J. E. 2001,
908:   \aj, 122, 2129
909: \bibitem[Israelian et al. (2004)]{israelian} Israelian, F., Ecuvillon, A., Rebolo, R., et al. 2004, A\&A, 421, 649
910: \bibitem[Izotov et al. (2006)]{iz05} Izotov, Y.I., Stasinsda, G., Meyenet, G., Guseva, N. G., \& Thuan, T. X. 2006, A\&A, 448, 955
911: \bibitem[Kauffmann et al.~(2003a)]{kauf03a} Kauffmann,
912:   F. et al. 2003, MNRAS, 341,33
913: \bibitem[Kauffmann et al.~(2003b)]{kauf03b} Kauffmann, G. et
914:   al. 2003,  MNRAS, 346, 1055
915: \bibitem[Kennicutt, Bresolin \& Garnett (2003)]{kbg} Kennicut,
916:   R. C., Bresolin, F., \& Garnett, D. R. 2003, ApJ, 591, 801
917: %\bibitem[Kennicutt (1998)]{ken98} Kennicutt,
918: %  R. C. 1998, ARA\&A, 36, 189
919: \bibitem[Kewley, Jansen \& Geller (2005)]{kjg} Kewley, L. J., Jansen
920:   R. A., \& Geller, M. J. 2005, PASP 117, 227
921: \bibitem[Kewley \& Dopita (2002)]{kd03}
922:   Kewley, L. J., \& Dopita, M. A. 2002, ApJS, 142, 35
923: \bibitem[Kewley et al.~(2001)]{kew01} Kewley, L. J., Dopita, M. A., Sutherland, R. S.,
924:   Heisler, C. A., \& Trevena, J. 2001, ApJ, 556,121
925: %\bibitem[Kobulnicky \& Skillman (1996)]{kobul96} Kobulnicky,
926: %  H. A. \& Skillman, E. D. 1996, ApJ, 471, 211
927: \bibitem[Kobulnicky, Kennicutt, \&  Pizagno
928:   (1999)]{kobul} Kobulnicky, H. A., Kennicutt, \& R. C, Pizagno, J. L.
929:   1999, ApJ, 514, 544
930: \bibitem[Kroupa (2001)]{kroupa} Kroupa P., 2001, MNRAS, 322,231
931: %\bibitem[Lehnert \& Heckman (1996)]{leh}
932: %  Lehnert,  M. D. \& Heckman, T. M. 1996, ApJ, 472, 546
933: %\bibitem[Lee et al (2006)]{Lee}, H., Skillman, E. D., Cannon,
934: %  J. M., Jackson, D. C., Gehrz, R. D., Polomski, E. F., \&
935: %  Woodward, C. E. 2006, ApJ, %%%%%%%%%%%%%%%%
936: 
937: \bibitem[Liang et al (2006)]{liang} Linag, Y. C., Yin, S. Y., Hammer, F., Deng, L., C., Flores, H., \& Zhang, B. 2006,
938:    ApJ, 652, 257
939: 
940: \bibitem[Maeder (2000)]{maeder} Maeder, A. 2000, New
941:   Astronomy Review, 44, 291
942: \bibitem[Martin et al. (2008)]{martin} Martin, C. D. et
943:   al. 2008  this volume
944: \bibitem[Matteucci (1986)]{mat86} Matteucci, F. 1986,
945:   MNRAS, 221 911
946: \bibitem[Matteucci \& Tornambe (1985)]{matt_t} Matteucci, F., Tornambe, A. 1985, A\&A, 142, 13
947: \bibitem[Matteucci \& Tosi (1985)]{mat85}Matteucci,
948:   F., \& Tosi, M. 1985, MNRAS 217 391
949: \bibitem[McGaugh (1991)]{mcgaugh} McGaugh, S. S. 1991,
950:   ApJ, 380, 140
951: %\bibitem[Miller \& Matthews (1972)]{mm} Miller,
952: %  J. S. \& Matthews, W. G. 1972, ApJ, 172, 591
953: \bibitem[Milliard et l.(1992)]{mil} Milliard B., Donas J.,
954:   Laget M., Armand C., Vuillemin A.  1992, A\&A, 257, 24
955: \bibitem[Morrissey et al.~(2008)]{pm} Morrissey, P. et
956:   al. 2008, this volume
957: \bibitem[Oey et al.~(2000)]{oey} Oey, M. S., Dopita,
958:  M. A., Shields, J. C., Smith, R. C. 2000, ApJS, 128, 511
959: \bibitem[Osterbrock (1989)]{oster} Osterbrock,
960:   D. E. 1989, Astrophysics of Gaseous Nebulae and Active
961:   Galactic Nuclei (Mill Valley:University Science Books)
962: \bibitem[Pagel et al.~(1979)]{pag79} Pagel, B. E. J.,
963:   Edmonds,  M. G., Blackwell, D. E., Chun, M. S., \& Smith,
964:   G. 1979,  MNRAS, 189, 95
965: \bibitem[Pagel et al.~(1992)]{pag92} Pagel,
966:   B. E. J., Simonson, E. A., Terlevich, R. J., \& Edmunds,
967:   M. G. 1992, MNRAS 255 325
968: %\bibitem[Pagel (1997)]{pagel97} Pagel, B. E. J. 1997 in Nucleosynthesis and Chemical Evolution of Galaxies. Cambridge Univ. Press, Cambridge
969: \bibitem[Peimbert et al (2006)]{peimbert} Peimbert, M., Peimbert, A., Esteban, C., 
970: Garcia-Rojas, J., Bresolin, F., Carigi, L., Ruiz, M.T., \& Lopez-Sanchez, A.R. 2006, RevMexAA, in press
971: 
972: \bibitem[Pettini et al.~(2000)]{petini} Pettini, M.,
973:   Steidel, C. C., Adelberger, D. L., \& Giavalisco, M. 2001,
974:   ApJ, 528, 96
975: \bibitem[Pettini \& Pagel (2004)]{pp04} Pettini, M., \& Pagel,
976:   B. E. J. 2004, MNRAS, 348, 59
977: \bibitem[Pier et al.~(2003)]{pier} Pier, J. R.,  Munn,
978:   J. A., Hindsley, R. B., Hennessy, G. S., Kent, S. M.,
979:   Lupton, R. H., \& Ivezic, Z. 2003, 125, 1559
980: %\bibitem[Rauch (2002)]{rauch} Rauch, T. 2002,
981: %  Rev. Mex. Astron. Astrofis. Conf. Ser. 12, 150
982: %\bibitem[Rich, M. R. et al. (2005)]{rich05} Rich, R. M. 2005,
983: %  ApJ, 619 107
984: \bibitem[Salim et al.~(2005)]{ss05} Salim, S. et al. 2005, ApJ, 619, 39
985: \bibitem[Salim et al.~(2008)]{ss07} Salim, S. et al. 2008, this volume
986: %\bibitem[Scalo (1986)]{scalo} Scalo, J. M. 1986. {\it
987: %  Fundam. Cosm. Phys.} 11:1-278 
988: %\bibitem[Schlegel Finkbeiner \& Davis  (1998)]{schlegel} Schlegel, D. J., Finkbeiner, D. P., \&
989: %  Davis, M. 1998, ApJ, 500, 525
990: \bibitem[Scully et al. (1996)]{scully} Scully, S., Casse, M., Olive, K. A., \& Vangioni-Flam, E. 1996, ApJ, 462, 960
991: \bibitem[Seaton (1978)]{seaton} Seaton, M. J. 1979,
992:   MNRAS 187, 785
993: \bibitem[Seibert et al.~(2005)]{sbrt05} Seibert, M. et
994:   al. 2005, ApJ, 619, 23
995: \bibitem[Shaw \& Dufour (1995)]{fivel} Shaw, R. A., \& Dufour, R. J. 1995, PASP, 107, 896 
996: \bibitem[Silk (1995)]{silk} Silk, J. 1995, ApJ, 438, L41
997: \bibitem[Smith et al.~(2002)]{smith02} Smith, J. A. et
998:   al. 2002, \aj, 123, 2121
999: \bibitem[Spite et al. (2005)]{spite} Spite, M., Cayrel, R., Plez, B., et al. 2005, A\&A, 430, 655 (S05)
1000: \bibitem[Stasinka (1990)]{stasinka} Stasinka, G. 1990,
1001:   A\&AS, 83, 501
1002: \bibitem[Stoughton et al. (2002)]{stoughton} Stoughton,
1003:   C. et al. 2002, \aj, 123, 485
1004: %\bibitem[Terlevich, Caldwell, \& Bower,
1005: %  (2001)]{terlevich} Terlevich, A. I., Caldwell, N., \&
1006:   Bower, R. G. 2001, MNRAS,  326, 1547
1007: \bibitem[Thurston, Edmunds \& Henry
1008:   (1996)]{thurston}Thurston, T. R., Edmunds, M. G.,
1009:   Henry, R. B. C. 1996, MNRAS, 283, 990
1010: \bibitem[Tremonti et al.(2004)~]{trem05}  Tremonti,
1011:   C. A. et al. 2004, ApJ, 613, 898
1012: %\bibitem[van Zee, Haynes, \& Salzer (1997)]{vzee97} van Zee, L., Haynes,
1013: %  M. P. \& Salzer, J.J. 1997, AJ, 114, 2479
1014: %\bibitem[van Zee, Salzer, \& Haynes (1998)]{vzee98} van Zee, L., Salzer, J.J., \& Haynes,
1015: %  M. P. 1998, ApJ, 497, 1
1016: 
1017: \bibitem[van Zee \& Haynes (2006)]{vzee06} van Zee, L., \&
1018:   Haynes, M. P. 2006, ApJ, 636, 214
1019: 
1020: \bibitem[Vila-Costas \& Edmunds (1993)]{villa93}
1021:   Vila-Costas, M. V. \& Edmunds M. G. 1992, MNRAS, 265, 199
1022: \bibitem[Woosley \& Weaver (1995)]{woos} Woosley, S. E., \& Weaver, T. A.  1995, ApJS, 101, 181
1023: \bibitem[York et al. (2000)]{york} York, D. G., Adelman, J., Anderson, J. E., Jr., et al. 2000, AJ, 120, 1579 
1024: \end{thebibliography}
1025: \clearpage
1026: 
1027: 
1028: 
1029: \begin{figure}
1030: \plotone{f1.eps}
1031: %\plotone{/Users/rmallery/Desktop/Work/papers/Nitro_GLX_SDSS_MS/f1.ps}
1032: \caption{Comparison of abundances from three different diagnostics: O3N2, M91, and T04 as a function of stellar mass.
1033:         The data points are converted into a 75 by 75 pixel image. The mean specific SFR value of the points in each pixel is
1034:         calculated and byte-scaled into true color. The difference between T04 and the other two diagnostics shows a dependence on mass
1035:         since T04 estimates an increasingly larger metallicity at higher stellar masses. 
1036:         }
1037: \end{figure}
1038: 
1039: 
1040: 
1041: %\begin{figure}
1042: %\plotone{f2.eps}
1043: %\plotone{/Users/rmallery/Desktop/Work/papers/Nitro_GLX_SDSS_MS/f5.ps}
1044: %\caption{Comparison of N/O ratios calculated with the strong-line method versus the {\it T$_e$} method for the 33 galaxies with 3$\sigma$ detections
1045: %        of the temperature sensitive emission line [OIII]4363. The strong line calculation is calculated
1046: %         by estimating the average NII region electron temperature with $R_{23}$ using the relation of \citet{thurston}. The temperatures are then used
1047: %         to derive the ratio of N/O from the strong line calibration of \citet{pag92}. The mean difference between the strong line and the {\it T$_e$} method
1048: %         is  $0.1$ dex with a scatter of $0.07$ dex.}  
1049: %\end{figure}
1050: %\clearpage
1051: 
1052: 
1053: \begin{figure}
1054: \plotone{f2.eps}
1055: %\plotone{/Users/rmallery/Desktop/Work/papers/Nitro_GLX_SDSS_MS/f2.ps}
1056: \caption{N/O versus O/H for three different methods of  abundance determination. The points are binned into a 75 by 75 pixel image, with the 
1057:           mean value of specific star formation rate (SFR/M$_*$) calculated from the points in each pixel and byte scaled into a true color representation.
1058:           The specific star formation rate is an indicator of the star formation history of the galaxy.
1059:           The plot shows the general trend that galaxies with similar metallicities have lower N/O ratios for larger values of SFR/M$_*$.
1060:           This trend supports the time delay scenario where the the bulk of the oxygen is released from short lived massive stars, 
1061:               and the release of the bulk of the nitrogen from longer lived intermediate mass stars.}
1062: \end{figure}
1063: \clearpage
1064: 
1065: \begin{figure}
1066: \plotone{f3.eps}
1067: %\plotone{/Users/rmallery/Desktop/Work/papers/Nitro_GLX_SDSS_MS/f3.ps}
1068: \caption{2D Histogram of N/O versus O/H. The red points represent the mean N/O values for objects with SFR/M$_*< -10.1$ taken for metallicity increments of
1069:           $0.1$ dex. The green points 
1070:      represent objects with $ -10.1<$SFR/M$_*< -9.1$, and the blue data points represent galaxies with SFR/M$_*> -.9.1$. The abundance methods of O3N2, M91 and 
1071:      T04 all show that the starbursts (having values of  SFR/M$_*> -10.1$) have lower values of N/O than galaxies of similar metallicity
1072:      that are currently not forming as large a fraction of their stellar mass.}
1073: \end{figure}
1074: \clearpage
1075: 
1076: \begin{figure}
1077: \plotone{f4.eps}
1078: %\plotone{/Users/rmallery/Desktop/Work/papers/Nitro_GLX_SDSS_MS/f4.ps}
1079: \caption{Difference of  mean N/O ratios between three sub-samples of specific star formation rate and the mean N/O of the entire sub-sample taken
1080:         for metallicity increments of $0.1$dex. 
1081:          The red points represent the mean N/O values for objects with SFR/M$_*< -10.1$, the green points 
1082:      represent objects with $-10.1<$SFR/M$_*< -9.1$, and the blue data points represent galaxies with SFR/M$_*> -9.1$. The abundance methods of O3N2, M91 and 
1083:      T04 all show that galaxies with the lowest specific star formation rates  (having values of SFR/M$_* < -10.1$) have higher values of N/O than galaxies of 
1084:       similar metallicity with  log SFR/M$_* > -10.1$. At intermediate metallicities the most extreme starbursts (log SFR/M$_* > -9.1$) on average have slightly 
1085:        lower N/O  ratios than
1086:       galaxies with average specific star formation rates ($-10.1<$log SFR/M$_*<-9.1$ by $0.02$ dex presumably due to oxygen released by the high mass stars formed
1087:        in the starburst}.
1088: \end{figure}
1089: \clearpage
1090: 
1091: 
1092: \begin{figure}
1093: \plotone{f5.eps}
1094: %\plotone{/Users/rmallery/Desktop/Work/papers/Nitro_GLX_SDSS_MS/f5.ps}
1095: \caption{N/O versus O/H scaled with Log M$_{*}$. This shows that  M$_{*}$ increases with increasing metallicity, and tends to have little dependence on N/O at a given
1096:           metallicity.}
1097: \end{figure}
1098: \clearpage
1099: 
1100: \begin{figure}
1101: \plotone{f6.eps}
1102: %\plotone{/Users/rmallery/Desktop/Work/papers/Nitro_GLX_SDSS_MS/f6.ps}
1103: \caption{N/O versus O/H, scaled with {\it g-r}. This plot shows that the {\it g-r} color of galaxies is redder for galaxies with higher metallicities. 
1104:          Also for  galaxies with similar metallicities, those galaxies with larger values of N/O have a redder color. This is because redder galaxies have increasingly
1105:          declining SFRs, where the intermediate mass stars from previous star formation events have released nitrogen into the ISM. This confirms the result of
1106:          \citet{vzee06} who found a similar trend for dwarf galaxies. }
1107: \end{figure}
1108: \clearpage
1109: 
1110: \begin{figure}
1111: \plotone{f7.eps}
1112: %\plotone{/Users/rmallery/Desktop/Work/papers/Nitro_GLX_SDSS_MS/f7.ps}
1113: \caption{N/O versus O/H  for abundances calculated using the {\it $T_e$} method. Overall the galaxies with $\log$ N/O$<-1.5$ on average tend to be slightly bluer, with 
1114:          higher equivalent widths, and with specific star formation rates $0.12$ dex higher 
1115:           than galaxies with $\log$ N/O$>-1.5$. The {\it $T_e$} oxygen abundances show little correlation with 
1116:          stellar mass, but this is probably due to small sample size and errors in the derived abundances. A larger sample size of galaxies with abundances 
1117:         derived from the  {\it $T_e$} method with a greater range of equivalent widths, stellar masses, and specific star formation rates, are needed to confirm
1118:          the results from O3N2, M91, and T04.}
1119: \end{figure}
1120: \clearpage
1121: 
1122: 
1123: 
1124: 
1125: \begin{deluxetable}{lll}
1126: \tabletypesize{\scriptsize}
1127: \tablecaption{Average Galaxy Parameter Errors}
1128: \tablewidth{0pc}
1129: \tablecolumns{3}
1130: \tablehead{
1131: \colhead{Parameter} &\colhead{$<value>$}  &\colhead{$<     \sigma>$}
1132: }
1133: \startdata
1134: log M$_*$   &10.10  &0.075 \\
1135: $<log SFR>_{100 Myr}$   &0.48 &0.20\\
1136: $<\log$ SFR/M$_*>_{100 Myr}$  &-9.6 &0.19\\ 
1137: log b                  &-0.24   &0.22\\
1138: A$_{FUV}$             &2.40 &0.56\\
1139: A$_{NUV}$               &1.78  &0.43\\
1140: \enddata
1141: %\tablenotetext{a}{  ~ref{table-1}}
1142: 
1143: \end{deluxetable}
1144: 
1145: 
1146: 
1147: 
1148: 
1149: \begin{deluxetable}{lrllrrlrl}
1150: \tabletypesize{\scriptsize}
1151: \tablecaption{${\it T_e}$ Abundances}
1152: \tablewidth{0pc}
1153: \tablecolumns{9}
1154: \tablehead{ \colhead{$\alpha$J2000} &\colhead{dJ2000} &\colhead{$12+\log O/H$}  &\colhead{log N/O}
1155:     &\colhead{log $SFR/M_*$}  &\colhead{$M_*$}   &\colhead{$F_{BURST}~ ^{1}$}  &\colhead{{\it g-r}} &\colhead{$\log EW_{H\alpha}$} 
1156:     }
1157: \startdata
1158: 20$^{h}$40$^{m}$18.047$^s$ & 01$^o$03$^{\prime}$24$^{\prime\prime}$.59 &8.31 $\pm$0.023 &-1.43 $\pm$0.047 &-8.19 $\pm$0.087 & 8.28 $\pm$0.057 &0.00$^{0.65}_{0.00}$ &-0.13 &2.355\\
1159: 20$^{h}$52$^{m}$51.797$^s$ & 00$^o$16$^{\prime}$26$^{\prime\prime}$.00 &8.16 $\pm$0.028 &-1.38 $\pm$0.057 &-8.89 $\pm$0.170 & 9.54 $\pm$0.115 &0.10$^{0.15}_{0.00}$ & 0.31 &2.091\\
1160: 21$^{h}$18$^{m}$29.846$^s$ & 00$^o$30$^{\prime}$59$^{\prime\prime}$.55 &8.18 $\pm$0.017 &-1.56 $\pm$0.049 &-9.16 $\pm$0.178 & 9.26 $\pm$0.115 &0.00$^{0.15}_{0.00}$ & 0.39 &2.506\\
1161: 22$^{h}$07$^{m}$07.888$^s$ & 00$^o$46$^{\prime}$58$^{\prime\prime}$.78 &8.10 $\pm$0.030 &-1.62 $\pm$0.086 &-8.39 $\pm$0.150 & 8.11 $\pm$0.082 &0.00$^{0.51}_{0.00}$ & 0.00 &2.216\\
1162: 22$^{h}$12$^{m}$23.328$^s$ & 00$^o$03$^{\prime}$39$^{\prime\prime}$.86 &8.24 $\pm$0.031 &-1.40 $\pm$0.062 &-9.06 $\pm$0.247 & 9.41 $\pm$0.128 &0.05$^{0.32}_{0.00}$ & 0.26 &2.354\\
1163: 21$^{h}$50$^{m}$29.868$^s$ & 00$^o$32$^{\prime}$01$^{\prime\prime}$.26 &8.24 $\pm$0.032 &-1.53 $\pm$0.070 &-8.24 $\pm$0.173 & 8.19 $\pm$0.105 &0.00$^{0.56}_{0.00}$ &-0.09 &2.349\\
1164: 20$^{h}$46$^{m}$56.140$^s$ & 00$^o$50$^{\prime}$37$^{\prime\prime}$.63 &8.27 $\pm$0.032 &-1.66 $\pm$0.070 &-8.34 $\pm$0.192 & 8.90 $\pm$0.140 &0.38$^{0.73}_{0.00}$ & 0.10 &2.279\\
1165: 10$^{h}$53$^{m}$42.546$^s$ & 00$^o$09$^{\prime}$45$^{\prime\prime}$.13 &8.19 $\pm$0.024 &-1.55 $\pm$0.053 &-8.66 $\pm$0.188 & 9.41 $\pm$0.113 &0.18$^{0.47}_{0.07}$ & 0.14 &2.258\\
1166: 17$^{h}$09$^{m}$22.632$^s$ & 61$^o$48$^{\prime}$51$^{\prime\prime}$.25 &8.36 $\pm$0.026 &-1.45 $\pm$0.048 &-8.47 $\pm$0.155 & 9.52 $\pm$0.118 &0.25$^{0.63}_{0.15}$ & 0.22 &2.440\\
1167: 00$^{h}$53$^{m}$00.523$^s$ & 15$^o$01$^{\prime}$29$^{\prime\prime}$.73 &8.19 $\pm$0.021 &-1.58 $\pm$0.047 &-8.28 $\pm$0.153 & 8.36 $\pm$0.062 &0.00$^{0.77}_{0.00}$ & 0.00 &2.183\\
1168: 08$^{h}$01$^{m}$43.632$^s$ & 44$^o$54$^{\prime}$58$^{\prime\prime}$.41 &8.37 $\pm$0.025 &-1.46 $\pm$0.052 &-8.73 $\pm$0.092 & 9.16 $\pm$0.085 &0.10$^{0.24}_{0.00}$ & 0.04 &1.985\\
1169: 08$^{h}$20$^{m}$01.714$^s$ & 50$^o$50$^{\prime}$39$^{\prime\prime}$.20 &8.35 $\pm$0.020 &-1.59 $\pm$0.045 &-8.94 $\pm$0.120 & 9.82 $\pm$0.053 &0.00$^{0.14}_{0.00}$ & 0.43 &2.537\\
1170: 08$^{h}$47$^{m}$03.007$^s$ & 54$^o$50$^{\prime}$39$^{\prime\prime}$.45 &8.22 $\pm$0.019 &-1.53 $\pm$0.060 &-8.30 $\pm$0.162 & 9.28 $\pm$0.125 &0.41$^{0.43}_{0.09}$ & 0.15 &2.356\\
1171: 03$^{h}$05$^{m}$39.705$^s$ &-08$^o$39$^{\prime}$05$^{\prime\prime}$.24 &8.20 $\pm$0.027 &-1.55 $\pm$0.083 &-8.36 $\pm$0.163 & 8.65 $\pm$0.092 &0.10$^{0.51}_{0.00}$ &-0.06 &2.279\\
1172: 12$^{h}$05$^{m}$14.725$^s$ & 66$^o$16$^{\prime}$57$^{\prime\prime}$.80 &8.32 $\pm$0.030 &-1.50 $\pm$0.058 &-9.34 $\pm$0.072 & 9.61 $\pm$0.047 &0.00$^{0.03}_{0.00}$ & 0.44 &2.346\\
1173: 10$^{h}$23$^{m}$19.567$^s$ & 02$^o$49$^{\prime}$41$^{\prime\prime}$.53 &8.08 $\pm$0.031 &-1.35 $\pm$0.059 &-8.39 $\pm$0.243 & 9.05 $\pm$0.140 &0.05$^{0.31}_{0.00}$ & 0.16 &2.218\\
1174: 11$^{h}$36$^{m}$55.796$^s$ & 03$^o$33$^{\prime}$33$^{\prime\prime}$.40 &8.34 $\pm$0.024 &-1.54 $\pm$0.047 &-9.00 $\pm$0.158 & 9.73 $\pm$0.117 &0.06$^{0.19}_{0.00}$ & 0.26 &2.422\\
1175: 08$^{h}$39$^{m}$14.949$^s$ & 48$^o$15$^{\prime}$18$^{\prime\prime}$.24 &8.17 $\pm$0.027 &-1.47 $\pm$0.053 &-8.51 $\pm$0.217 & 8.49 $\pm$0.112 &0.00$^{0.37}_{0.00}$ & 0.13 &2.136\\
1176: 09$^{h}$46$^{m}$30.590$^s$ & 55$^o$35$^{\prime}$41$^{\prime\prime}$.81 &8.23 $\pm$0.031 &-1.53 $\pm$0.062 &-8.58 $\pm$0.140 & 8.76 $\pm$0.075 &0.00$^{0.21}_{0.00}$ & 0.08 &2.084\\
1177: 14$^{h}$05$^{m}$01.154$^s$ & 04$^o$31$^{\prime}$26$^{\prime\prime}$.13 &8.46 $\pm$0.027 &-1.74 $\pm$0.054 &-8.23 $\pm$0.105 & 8.00 $\pm$0.035 &0.00$^{0.65}_{0.00}$ &-0.06 &2.297\\
1178: 14$^{h}$36$^{m}$48.204$^s$ & 04$^o$02$^{\prime}$59$^{\prime\prime}$.92 &8.17 $\pm$0.031 &-1.34 $\pm$0.060 &-9.44 $\pm$0.070 & 9.06 $\pm$0.082 &0.00$^{0.02}_{0.00}$ & 0.23 &2.077\\
1179: 14$^{h}$46$^{m}$10.316$^s$ & 03$^o$39$^{\prime}$21$^{\prime\prime}$.55 &8.31 $\pm$0.016 &-1.46 $\pm$0.032 &-8.71 $\pm$0.162 & 9.75 $\pm$0.100 &0.16$^{0.26}_{0.06}$ & 0.28 &2.437\\
1180: 14$^{h}$54$^{m}$24.609$^s$ & 03$^o$59$^{\prime}$25$^{\prime\prime}$.20 &8.29 $\pm$0.020 &-1.49 $\pm$0.041 &-8.81 $\pm$0.185 & 9.71 $\pm$0.075 &0.00$^{0.47}_{0.00}$ & 0.36 &2.543\\
1181: 00$^{h}$52$^{m}$49.794$^s$ &-08$^o$41$^{\prime}$33$^{\prime\prime}$.93 &8.10 $\pm$0.034 &-1.49 $\pm$0.070 &-8.51 $\pm$0.202 & 8.48 $\pm$0.145 &0.00$^{0.46}_{0.00}$ & 0.02 &2.163\\
1182: 01$^{h}$38$^{m}$44.917$^s$ &-08$^o$35$^{\prime}$40$^{\prime\prime}$.69 &8.18 $\pm$0.017 &-1.62 $\pm$0.045 &-8.24 $\pm$0.235 & 8.60 $\pm$0.033 &0.42$^{0.68}_{0.00}$ &-0.09 &2.303\\
1183: 01$^{h}$47$^{m}$21.680$^s$ &-09$^o$16$^{\prime}$46$^{\prime\prime}$.23 &8.31 $\pm$0.018 &-1.70 $\pm$0.050 &-9.37 $\pm$0.145 & 9.68 $\pm$0.115 &0.03$^{0.12}_{0.02}$ & 0.30 &2.642\\
1184: 02$^{h}$03$^{m}$56.913$^s$ &-08$^o$07$^{\prime}$58$^{\prime\prime}$.48 &8.37 $\pm$0.018 &-1.43 $\pm$0.036 &-8.43 $\pm$0.162 & 9.48 $\pm$0.090 &0.25$^{0.52}_{0.00}$ & 0.21 &2.414\\
1185: 22$^{h}$58$^{m}$33.743$^s$ & 00$^o$56$^{\prime}$30$^{\prime\prime}$.53 &8.13 $\pm$0.032 &-1.39 $\pm$0.081 &-8.65 $\pm$0.270 & 9.17 $\pm$0.193 &0.17$^{0.55}_{0.00}$ & 0.08 &2.377\\
1186: 23$^{h}$29$^{m}$32.117$^s$ & 00$^o$34$^{\prime}$26$^{\prime\prime}$.91 &8.32 $\pm$0.029 &-1.63 $\pm$0.079 &-9.15 $\pm$0.065 & 9.23 $\pm$0.100 &0.00$^{0.06}_{0.00}$ & 0.39 &2.382\\
1187: 22$^{h}$53$^{m}$56.829$^s$ & 10$^o$13$^{\prime}$00$^{\prime\prime}$.29 &8.31 $\pm$0.031 &-1.46 $\pm$0.060 &-8.88 $\pm$0.092 &10.04 $\pm$0.108 &0.10$^{0.10}_{0.01}$ & 0.41 &2.252\\
1188: 10$^{h}$21$^{m}$32.505$^s$ & 61$^o$44$^{\prime}$04$^{\prime\prime}$.52 &8.16 $\pm$0.023 &-1.48 $\pm$0.056 &-9.08 $\pm$0.207 & 9.36 $\pm$0.123 &0.06$^{0.22}_{0.00}$ & 0.16 &2.294\\
1189: 08$^{h}$20$^{m}$10.558$^s$ & 37$^o$43$^{\prime}$54$^{\prime\prime}$.34 &8.11 $\pm$0.026 &-1.66 $\pm$0.060 &-8.60 $\pm$0.162 & 9.09 $\pm$0.077 &0.00$^{0.38}_{0.00}$ & 0.13 &2.022\\
1190: 21$^{h}$19$^{m}$58.308$^s$ & 00$^o$52$^{\prime}$33$^{\prime\prime}$.52 &8.26 $\pm$0.016 &-1.55 $\pm$0.046 &-9.00 $\pm$0.435 & 8.52 $\pm$0.015 &0.07$^{0.07}_{0.00}$ &-0.06 &2.375
1191: \enddata
1192: \tablenotetext{1}{The superscripts for $F_{BURST}$ (Fraction of stellar mass formed in starbursts over the last 100Myr) list the 97.5 percentile values, and 
1193:        the suscripts list the 2.5 percentile values.}
1194: \end{deluxetable}
1195: 
1196: \begin{deluxetable}{llrlrrr}
1197: \tabletypesize{\scriptsize}
1198: \tablecaption{O3N2 mean N/O}
1199: \tablewidth{0pc}
1200: \tablecolumns{7}
1201: \tablehead{ \colhead{$12+\log O/H$} &\colhead{Sub-sample$^1$} &\colhead{\#of galaxies} 
1202:     &\colhead{$\log N/O_{mean}$} &\colhead{$\log SFR/M_{_*mean}$}  &\colhead{$\log M_{_*mean}$}
1203:     &\colhead{${\it g-r}_{mean}$}
1204:     }
1205: \startdata
1206:     &  a&   47 & -1.06 $\pm$0.013 &-10.49 $\pm$0.055 &10.14 $\pm$0.081 &0.69 $\pm$0.023\\
1207: 8.55&  b&  873 & -1.13 $\pm$0.003 & -9.49 $\pm$0.008 & 9.74 $\pm$0.011 &0.43 $\pm$0.003\\
1208:     &  c&  149 & -1.10 $\pm$0.005 & -8.88 $\pm$0.018 & 9.88 $\pm$0.029 &0.36 $\pm$0.007\\ \\
1209: 
1210:     &  a&  295 & -0.99 $\pm$0.006 &-10.34 $\pm$0.014 &10.27 $\pm$0.025 &0.71 $\pm$0.006\\
1211: 8.65&  b& 1638 & -1.04 $\pm$0.002 & -9.62 $\pm$0.006 &10.05 $\pm$0.009 &0.52 $\pm$0.002\\
1212:     &  c&  132 & -1.04 $\pm$0.006 & -8.89 $\pm$0.020 &10.06 $\pm$0.028 &0.43 $\pm$0.008\\\\
1213: 
1214:     &  a&  809 & -0.89 $\pm$0.004 &-10.33 $\pm$0.008 &10.53 $\pm$0.013 &0.74 $\pm$0.004\\
1215: 8.75&  b& 2459 & -0.94 $\pm$0.002 & -9.71 $\pm$0.005 &10.38 $\pm$0.007 &0.60 $\pm$0.002\\
1216:     &  c&  109 & -0.93 $\pm$0.008 & -8.88 $\pm$0.024 &10.48 $\pm$0.034 &0.55 $\pm$0.011\\\\
1217: 
1218:     &  a&  197 & -0.81 $\pm$0.008 &-10.34 $\pm$0.021 &10.58 $\pm$0.023 &0.74 $\pm$0.007\\
1219: 8.85&  b&  637 & -0.85 $\pm$0.003 & -9.68 $\pm$0.009 &10.50 $\pm$0.014 &0.63 $\pm$0.004\\
1220:     &  c&   36 & -0.85 $\pm$0.015 & -8.92 $\pm$0.030 &10.48 $\pm$0.050 &0.59 $\pm$0.015\\
1221: \enddata
1222:   \tablenotetext{1}{Sub-sample a: galaxies with  SFR/M$_* <-10.1$,   Sub-sample b: galaxies with $-10.1<$SFR/M$_* < -9.1$, Sub-sample c galaxies with SFR/M$_*>$-9.1}
1223: \end{deluxetable}
1224: 
1225: 
1226: \begin{deluxetable}{llccccc}
1227: \tabletypesize{\scriptsize}
1228: \tablecaption{M91 mean N/O}
1229: \tablewidth{0pc}
1230: \tablecolumns{7}
1231: \tablehead{ \colhead{12$+$log O/H} &\colhead{Sub-sample$^1$}  &\colhead{\#of galaxies} 
1232:      &\colhead{log N/O$_{mean}$} &\colhead{log SFR/M$_{*mean}$}  &\colhead{log M$_{*mean}$}
1233:    &\colhead{{\it g-r}$_{mean}$}
1234:    }
1235: \startdata
1236:      &  a &   64&-1.10 $\pm$0.009 &-10.48 $\pm$0.044 &10.12 $\pm$0.085 &0.69 $\pm$0.024 \\
1237:  8.65&  b &  733&-1.18 $\pm$0.003 & -9.47 $\pm$0.009 & 9.62 $\pm$0.017 &0.42 $\pm$0.004\\
1238:      &  c &  229&-1.21 $\pm$0.005 & -8.85 $\pm$0.016 & 9.58 $\pm$0.030 &0.31 $\pm$0.007\\\\
1239:     
1240:      &  a &  169&-1.02 $\pm$0.006 &-10.40 $\pm$0.024 &10.27 $\pm$0.036 &0.71 $\pm$0.009\\
1241:  8.75&  b & 1228&-1.09 $\pm$0.002 & -9.55 $\pm$0.007 & 9.92 $\pm$0.012 &0.48 $\pm$0.004\\
1242:      &  c &  191&-1.10 $\pm$0.004 & -8.88 $\pm$0.016 & 9.92 $\pm$0.027 &0.37 $\pm$0.008\\\\
1243: 
1244:      &  a &  360&-0.96 $\pm$0.004 &-10.33 $\pm$0.011 &10.37 $\pm$0.023 &0.72 $\pm$0.005\\
1245:  8.85&  b & 1733&-1.01 $\pm$0.001 & -9.64 $\pm$0.006 &10.17 $\pm$0.010 &0.55 $\pm$0.003\\
1246:      &  c &  126&-1.00 $\pm$0.009 & -8.84 $\pm$0.024 &10.13 $\pm$0.041 &0.46 $\pm$0.010\\\\
1247: 
1248:      &  a &  597&-0.86 $\pm$0.003 &-10.33 $\pm$0.009 &10.55 $\pm$0.015 &0.74 $\pm$0.004\\
1249:  8.95&  b & 1839&-0.91 $\pm$0.002 & -9.70 $\pm$0.006 &10.39 $\pm$0.009 &0.60 $\pm$0.002\\
1250:      &  c &   89&-0.91 $\pm$0.007 & -8.89 $\pm$0.024 &10.38 $\pm$0.041 &0.53 $\pm$0.012\\
1251:  \enddata
1252:  \end{deluxetable}
1253: \tablenotetext{1}{Sub-sample a: galaxies with  SFR/M$_* <-10.1$,   Sub-sample b: galaxies with $-10.1<$SFR/M$_* < -9.1$, Sub-sample c galaxies with SFR/M$_*>$-9.1}
1254: 
1255: 
1256: 
1257:  \begin{deluxetable}{llccccc}
1258:  \tabletypesize{\scriptsize}
1259:  \tablecaption{T04 mean N/O}
1260:  \tablewidth{0pc}
1261:  \tablecolumns{7}
1262:  \tablehead{ \colhead{$12+\log O/H$} &\colhead{Sub-sample$^1$}  &\colhead{\#of galaxies} 
1263:    &\colhead{$\log N/O_{mean}$} &\colhead{$\log M_{*mean}$}  &\colhead{$log L_{FUV}/M_{*mean}$}
1264:      &\colhead{${\it g-r}_{mean}$}
1265:      }
1266:   \startdata
1267:      &  a&   35 &-1.10 $\pm$0.018 &-10.38 $\pm$0.045 & 9.96 $\pm$0.102 &0.64 $\pm$0.028\\
1268:  8.75&  b&  506 &-1.17 $\pm$0.003 & -9.47 $\pm$0.010 & 9.63 $\pm$0.014 &0.42 $\pm$0.004\\
1269:      &  c&  128 &-1.18 $\pm$0.006 & -8.85 $\pm$0.021 & 9.67 $\pm$0.029 &0.32 $\pm$0.008\\\\
1270: 
1271:      &  a&  110 &-1.03 $\pm$0.008 &-10.41 $\pm$0.032 &10.03 $\pm$0.042 &0.67 $\pm$0.011 \\
1272:  8.85&  b& 1060 &-1.09 $\pm$0.002 & -9.55 $\pm$0.007 & 9.84 $\pm$0.010 &0.47 $\pm$0.003\\
1273:      &  c&  154 &-1.11 $\pm$0.005 & -8.86 $\pm$0.019 & 9.90 $\pm$0.027 &0.37 $\pm$0.008\\\\
1274: 
1275:      &  a&  297 &-0.95 $\pm$0.006 &-10.34 $\pm$0.013 &10.31 $\pm$0.022 &0.71 $\pm$0.006\\
1276:  8.95&  b& 1313 &-1.01 $\pm$0.002 & -9.63 $\pm$0.007 &10.10 $\pm$0.009 &0.53 $\pm$0.003\\
1277:      &  c&  133 &-1.04 $\pm$0.005 & -8.86 $\pm$0.022 &10.06 $\pm$0.033 &0.42 $\pm$0.009\\\\
1278: 
1279:      &  a&  551 &-0.89 $\pm$0.005 &-10.32 $\pm$0.009 &10.51 $\pm$0.015 &0.74 $\pm$0.004\\
1280:  9.05&  b& 1874 &-0.94 $\pm$0.002 & -9.70 $\pm$0.006 &10.37 $\pm$0.008 &0.59 $\pm$0.002\\
1281:      &  c&   95 &-0.95 $\pm$0.006 & -8.87 $\pm$0.027 &10.36 $\pm$0.036 &0.51 $\pm$0.011\\\\
1282: 
1283:      &  a&  338 &-0.83 $\pm$0.005 &-10.35 $\pm$0.015 &10.68 $\pm$0.017 &0.76 $\pm$0.005\\
1284:  9.15&  b&  935 &-0.86 $\pm$0.003 & -9.71 $\pm$0.008 &10.60 $\pm$0.011 &0.65 $\pm$0.003\\
1285:      &  c&   56 &-0.86 $\pm$0.011 & -8.93 $\pm$0.024 &10.59 $\pm$0.043 &0.60 $\pm$0.015\\
1286:  \enddata
1287: \tablenotetext{1}{Sub-sample a: galaxies with  SFR/M$_* <-10.1$,   Sub-sample b: galaxies with $-10.1<$SFR/M$_* < -9.1$, Sub-sample c galaxies with SFR/M$_*>$-9.1}
1288:  \end{deluxetable}
1289: \end{document}
1290: 
1291: 
1292: 
1293: 
1294: 
1295: 
1296: 
1297: