Different Dependence of Narrow Hα Line Luminosity on Optical Continuum Luminosity between Star-forming Galaxies and Type 2 Active Galactic Nuclei: Globally Negative AGN Feedback in Local Type 2 AGN?

In this manuscript, clues are provided to support globally negative active galactic nuclei (AGN) feedback on star formation in the host galaxies of the local low-redshift Sloan Digital Sky Survey (SDSS) Type 2 AGN, based on the different dependence of narrow Hα line luminosity L Hα on optical continuum luminosity λ L cont between star-forming galaxies and Type 2 AGN. Through the measured L Hα and λ L cont in SDSS star-forming galaxies, there is a strong linear correlation between λ L cont and L Hα , accepted as a standard correlation without the effects of AGN activity. Meanwhile, considering the apparent contribution of AGN activity to the narrow Hα line emissions in Type 2 AGN, the correlation between λ L cont and L Hα in the SDSS Type 2 AGN leads to a statistically lower L Hα in Type 2 AGN than in star-forming galaxies, with a significance level higher than 5σ, even after considering necessary effects (including effects of host galaxy properties), leading to the accepted conclusion on the globally negative AGN feedback in the local Type 2 AGN. Meanwhile, the properties of Dn(4000) and Hδ A can provide indirect clues to support the globally negative AGN feedback in local Type 2 AGN, due to older stellar ages in Type 2 AGN. Moreover, it is interesting to expect more than 50% narrow Hα emissions globally suppressed in the host galaxies of Type 2 AGN relative to the star-forming galaxies. The results not only support globally negative AGN feedback in local Type 2 AGN, but also show further clues on the quantification of suppressions of star formation by the globally negative AGN feedback.


Introduction
Active galactic nuclei (AGN) feedback through galacticscale outflows plays a key role in galaxy evolution, leading to the tight connections between AGNs and host galaxies McNamara & Nulsen (2007), Fabian (2012), Gaspari et al. (2012), Kormendy & Ho (2013), Heckman & Best (2014), King & Pounds (2015), Tombesi et al. (2015), Baron et al. (2018), Muller-Sanchez et al. (2018), Russell et al. (2019), Kolokythas et al. (2020), Richard-Laferriere et al. (2020), Chadayammuri et al. (2021), Kawamuro et al. (2021), Smethurst et al. (2021), and Piotrowska et al. (2022).Both observational and theoretical results have shown clear impacts of either positive or negative AGN feedback on star formation in the host galaxies of AGN. Feruglio et al. (2010) reported observational evidence to support negative AGN feedback in the nearest quasar Mrk 231, due to the detected giant molecular outflow with a higher mass rate than the detected star formation rate (SFR) in the host galaxy, therefore halting star formation.Page et al. (2012) showed evidence to support negative AGN feedback because rapid star formations are common in AGN host galaxies but the most vigorous star formations cannot be observed in AGN with X-ray luminosities larger than 10 44 erg s −1 .Wylezalek & Zakamska (2016) reported evidence for the negative AGN feedback, based on the AGN with strong outflow signatures hosted in galaxies that are more quenched than galaxies with weaker outflow signatures.Comerford et al. (2020) also showed evidence to support the negative AGN feedback, based on inside-out quenching of star formations in radio-mode AGN host galaxies, which have older stellar populations through a sample of 406 AGN subdivided into radio-quiet and radio-mode AGN.Meanwhile, evidence to support positive AGN feedback can be found in the literature.Zubovas et al. (2013) discussed the positive AGN feedback in gas-rich phases by overcompressing cold dense gas.Zinn et al. (2013) showed positive AGN feedback because of the much higher SFRs in the AGN with pronounced radio jets than in the purely X-ray-selected ones.Shin et al. (2019) reported a positive feedback scenario in NGC 5728, due to higher SFRs in the encountering region where the ionized gas outflows encounter the star formation ring at 1 kpc radius.Mahoro et al. (2022) showed that there are no apparent signs of negative AGN feedback, after comparing host galaxy properties of far-infrared AGN and non-AGN green valley galaxies.Therefore, it is interesting to provide further clues on the contradictory effects of AGN feedback on star formations, through different methods, which is the main objective of this manuscript.
Optical spectroscopic properties of star-forming galaxies with no contribution from central AGN activity can be well applied to trace star-forming histories, leading to the strong dependence of the SFRs on the narrow Hα line luminosities (L Hα ) in galaxies as well discussed in Kennicutt et al. (1994), Pflamm-Altenburg et al. (2007), Ly et al. (2011), Kennicutt & Evans (2012), Madau & Dickinson (2014), Smit et al. (2016), Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence.Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.Sanchez (2020), andVilla-Velez et al. (2021): In this manuscript, among all the narrow emission line main galaxies in the Sloan Digital Sky Survey (SDSS), Data Release 16 (SDSS DR16; Ahumada et al. 2020), star-forming galaxies with high-quality narrow emission lines can be collected, based on the dividing line applied in the Baldwin-Phillips-Terlevich (BPT) diagram (Baldwin et al. 1981;Kauffmann et al. 2003a;Kauffmann et al. 2003;Kewley et al. 2006;Kewley et al. 2013;Kewley et al. 2019;Zhang et al. 2020;Zhang 2022a) through the narrow emission line flux ratios of [O III]λ5007Å to Hβ (O3HB) and of [N II]λ6583Å to Hα (N2HA), which will be discussed in Section 2. Based on the properties of continuum emissions and narrow Hα emissions of the star-forming galaxies with no contribution from AGN activity on the spectroscopic features, the dependence of narrow Hα line luminosity on optical continuum luminosity λL cont in the starforming galaxies can be used as a standard candle to check effects of AGN feedback on Type 2 AGN.Unlike star-forming galaxies, Type 2 AGN have optical spectroscopic narrow emission lines including the apparent contribution of central AGN activity.Based on the commonly accepted and constantly being improved unified model (Netzer 2015;Suh et al. 2019) of AGN, emissions from central accretion disks and central broad emission line regions are totally obscured by central dust torus (Davies et al. 2015;Netzer et al. 2016;Martinez-Paredes et al. 2017;Zhuang et al. 2018;Prieto et al. 2021;Zhang 2022b;Zhang 2023) in Type 2 AGN.Therefore, the narrow emission lines of Type 2 AGN have contributions from both central AGN activity and star formation; however, both the continuum emissions and the narrow absorption features in the host galaxies of Type 2 AGN have little contribution from central AGN activity, which is strongly supported by optical spectra (with emission line features being masked out) of Type 2 AGN described by pure stellar templates without the consideration of the contribution of AGN as discussed in Section 3. Hence, studying the properties of the narrow Hα emissions in a large sample of Type 2 AGN can provide clues to the effects of AGN feedback, after considering the effects of AGN activity on the observational narrow Hα line emissions but few effects on the observational continuum emissions in the host galaxies of Type 2 AGN.
The manuscript is organized as follows.Section 2 presents the data samples of star-forming galaxies and Type 2 AGN.Section 3 presents our main results and a discussion.Section 4 provides our summary and conclusions.In this manuscript, we have adopted the cosmological parameters of H 0 = 70 km s −1 Mpc −1 , Ω Λ = 0.7, and Ω m = 0.3.

Data Samples
All the low-redshift (z < 0.35) main galaxies with highquality spectra (median spectral signal-to-noise ratio (S/N) larger than 20) are first collected from SDSS DR16, through the SDSS provided SQL (structured query language) 1 Search tool by the following query SELECT S.plate, S.fiberid, S.mjd, S.z, S. snmedian, P.petroR50_u, P.petroR50err_u, P.pet-roR50_g, P.petroR50err_g, P.petroR50_r, P.petroR50err_r, P.pet-roR50_i, P.petroR50err_i, P.petroR50_z, P.petroR50err_z, P.pet-roR90_u, P.petroR90err_u, P.petroR90_g, P.petroR90err_g, P.pet-roR90_r, P.petroR90err_r, P.petroR90_i, P.petroR90err_i, P.pet-roR90_z, P.petroR90err_z, P.devab_u, P.devab_g, P.devab_r, P.devab_i, P.devab_z, M.mstellar_median, M.mstellar_err FROM SpecObjall as S JOIN PhotoObjAll as P ON S.bestobjid=P.objidJOIN stellarMassPCAWiscM11 as M ON M.spe-cobjid=S.specobjidWHERE S.class=''GALAXY'' and S.z < 0.35 and S.zwarning=0 and S.snmedian > 20 Here, the restriction z < 0.35 is applied, to ensure that the narrow Hα and [N II] emission lines are totally covered in the SDSS spectra, which will be used to classify the main galaxies through the BPT diagrams.Meanwhile, the properties of the inverse concentration (IC) parameter will be discussed in Section 3; therefore, the SDSS public database of "Photo-ObjAll" is also considered in the SQL query, in order to collect the corresponding photometric information.Detailed descriptions on the database "PhotoObjAll" containing full photometric catalog quantities and on the database "SpecObjall" containing all the spectroscopic information can be found online.2Furthermore, properties of the total stellar mass (the parameter mstellar_median and the corresponding uncertainty mstellar_err) will be discussed in Section 3; therefore, the SDSS public database of "stellarMassPCAWiscM11" as described in detail in Maraston & Stromback (2011), Chen et al. (2012) is also considered in the SQL query.
Before proceeding further, spectroscopic features of the collected main galaxies are carefully checked, in order to measure emission lines after subtractions of the simple stellar population (SSP) method determined stellar continuum.In this manuscript, the commonly applied SSP (Bruzual & Charlot 1993;Kauffmann et al. 2003b;Cid Fernandes et al. 2005;Cid Fernandes et al. 2013;Lopez Fernandez et al. 2016;Cappellari 2017;Werle et al. 2019) method is accepted to determine the contribution of stellar lights in the SDSS spectra, with the 39 SSP templates discussed in Bruzual & Charl1ot (1993) and Kauffmann et al. (2003b), which include the population age from 5 Myr to 12 Gyr with three solar metallicities (Z = 0.008, 0.05, 0.02).Through the Levenberg-Marquardt least-squares minimization method3 (Markwardt 2009), the sum of the strengthened, broadened, and shifted SSP templates can be applied to describe the SDSS spectrum with emission lines being masked out by line width of about 400 km s −1 at zero intensity, similar to what we have recently done in Zhang et al. (2019), Zhang (2022b), Zhang (2022c), andZhang (2023).The left panels of Figure 1 show examples of the SSP method that determined the best descriptions and the corresponding line spectrum of a Type 2 AGN SDSS 0533-51994-0031 (plate-mjd-fiberid).Here, the line spectrum is calculated by the SDSS spectrum minus the SSP method determined by the stellar continuum.
After the subtraction of the stellar continuum, emission lines can be described by Gaussian functions.Here, the narrow emission lines of the Hα, Hβ, [O III]λ4959, 5007Å doublet and the [N II]λ6548, 6583Å doublet are mainly considered, in order to classify SDSS main galaxies by properties of the narrow emission line flux ratios in the BPT diagram of O3HB versus N2HA.Each Gaussian component is applied to describe each narrow emission line, besides the [O III] doublet, which are described by core plus extended broad Gaussian components probably related to shifted wings (Greene & Ho 2005;Shen et al. 2011;Zhang 2021).Due to the few effects of broad emission lines, there are no severe restrictions on the model parameters of each Gaussian component, besides the flux ratio of the [O III] ([N II]) doublet fixed to the theoretical value of 3, and the emission line flux not smaller than zero.Then, the narrow emission lines can be measured through the Levenberg-Marquardt least-squares minimization method.As examples, the right panels of Figure 1 show the best-fitting results to the narrow emission lines in the line spectrum of the Type 2 AGN SDSS 0533-51994-0031 of which spectrum is shown in the left panels.Here, one point should be noted.As shown in the example in Figure 1, almost all the collected objects have a [N II] doublet and narrow Balmer lines that can be described without considering extended components.
Based on the measured reliable line parameters,4 the main galaxies with reliable measurements of the narrow emission lines are collected, based on the criteria that each measured line parameter is at least five times larger than its corresponding measured uncertainty and that the flux ratio (Balmer decrement) of the narrow Hα to the narrow Hβ is less than 6 to ignore effects of serious obscuration on the following results.
Then, the well-known BPT diagram of O3HB versus N2HA is applied to classify the main galaxies into star-forming galaxies and Type 2 AGN by the reported dividing lines in the literature (Kauffmann et al. 2003;Kewley et al. 2019;Zhang et al. 2020) As shown in the left panel of Figure 2, there are 19,351 main galaxies classified as star-forming galaxies in the BPT diagram lying below the dividing line shown as a solid purple line, and 4112 main galaxies classified as Type 2 AGN in the BPT diagram lying above the dividing line shown as a solid dark green line.Based on the measured line parameters, the median Balmer decrement is 4.10 in the star-forming galaxies and 4.18 in the Type 2 AGN, indicating no different obscuration effects on the following results.
Before ending this section, an additional point should be noted.Although the Type 2 AGN are collected from the SDSS pipeline classified main galaxies (with no expected broad emission lines), it is necessary to check whether there are any collected Type 2 AGN with probably broad emission lines and probably AGN continuum emissions included in the SDSS spectra.After subtractions of the stellar continuum, emission lines around Hα (rest wavelength from 6200-6800 Å) of all the collected 4112 Type 2 AGN have been remeasured by the narrow Gaussian functions applied to describe the narrow emission lines but by three additional broad Gaussian functions (second moment larger than 600 km s −1 ) applied to describe the probably broad Hα.Based on the criterion that the measured parameters are larger than five times their corresponding uncertainties in one of the three broad Gaussian components in the broad Hα, there are 285 Type 2 AGN with probably broad Hα.Therefore, there are 3836 (4121-285) Type 2 AGN in our final sample.

Main Results and Discussion
Based on the measured line parameters and the continuum emission features of the collected 19,351 star-forming galaxies, there is a strong linear correlation between the narrow Hα line luminosity (L Hα ) and the continuum luminosity at 5100 Å (λL cont ) with a Spearman's rank correlation coefficient of 0.88 (P null < 10 −15 ).In this manuscript, although only objects are collected with flux ratio smaller than 6 of narrow Hα to narrow Hβ, the L Hα , and λL cont have been reddening corrected for the objects with flux ratios of narrow Hα to narrow Hβ larger than 3.1, accepted 3.1 as the intrinsic flux ratio of narrow Hα to narrow Hβ.After considering the uncertainties in both coordinates, the best-fitting results shown in the left panel of Figure 3 are determined through the least trimmed squares (LTS) robust technique (Cappellari et al. 2013;Mahdi & Mohammad 2017).
The strong linear correlation in the star-forming galaxies can be used as a standard candle to check effects of AGN feedback, by comparing the correlations of L Hα versus λL cont between starforming galaxies and Type 2 AGN.
Based on the measured parameters of the Type 2 AGN, there is also a strong linear correlation between L Hα and λL cont , with a Spearman's rank correlation coefficient of 0.80 (P null < 10 −15 ), also shown in the left panel of Figure 3. Here, L Hα and λL cont are the reddening corrected values in the Type 2 AGN.After considering the uncertainties in both coordinates, the best-fitting results in the Type 2 AGN are determined by the LTS technique.
It is clear that the linear correlations have much different intercepts but similar slopes between Type 2 AGN and starforming galaxies.
The expected L Hα are statistically smaller for the given continuum luminosities in Type 2 AGN than in star-forming galaxies.The right panel of Figure 3 shows the distributions of the luminosity ratio R CL of λL cont to L Hα .The median values of R log CL ( ) are about 2.39 ± 0.29 and 2.99 ± 0.34 in the starforming galaxies and the Type 2 AGN, respectively.N2HA for all the narrow emission line main galaxies collected from the SDSS DR16, with the collected 4112 Type 2 AGN lying above the dividing line shown as a solid dark green line and the collected 19,351 star-forming galaxies lying below the dividing line shown as the solid purple line.The right panel shows the dependence of L Hα on the parameter O3HB.In the right panel, the contour filled with different shades of green represents the results in the Type 2 AGN, and the contour with contour levels in reddish lines represents the results in the star-forming galaxies.In each panel, the color bar is shown in the top region to represent the corresponding number densities related to different colors. ).The student's t-test technique can be applied to confirm that the star-forming galaxies and the Type 2 AGN have different mean values of R log CL ( )with a significance level higher than 5σ.Therefore, the different correlations between λL cont and L Hα can be confirmed in the star-forming galaxies and the Type 2 AGN.
Before proceeding further, the effects of the S/N of SDSS spectra are simply discussed as follows on the results shown in Figure 3 for the star-forming galaxies with S/N > 20 and the Type 2 AGN with S/N > 20.Among the star-forming galaxies and the Type 2 AGN, the 2957 star-forming galaxies with S/N > 30 and the 1060 Type 2 AGN with S/N > 30 are collected to re-check the correlation between L Hα versus λL cont , shown in Figure 4.The linear correlations can be confirmed with Spearman rank correlation coefficients of about 0.84 (P null < 10 −15 ) and 0.79 (P null < 10 −15 ) for the starforming galaxies with S/N > 30 and for the Type 2 AGN with S/N > 30, respectively.And the best-fitting results can be described as The median values and the corresponding standard deviations (as uncertainties) of R log CL ( )are about 2.37 ± 0.31 and 3.04 ± 0.35 in the 2957 star-forming galaxies with S/N > 30 and in the 1060 Type 2 AGN with S/N > 30, respectively.The student's t-test technique can be applied to confirm the different mean values of R log CL ( ) with a significance level higher than 5σ.The results similar to those shown in Figure 3 strongly indicate few effects of S/N on our final results.Therefore, there is no further discussion on the effects of the S/N.
Considering the strong connections between star-forming properties and narrow Hα line luminosities, the effects of AGN feedback on the narrow Hα line luminosity in Type 2 AGN can be expected.If there was positive AGN feedback on star formation, statistically stronger narrow Hα emissions could be expected in Type 2 AGN than in the star-forming galaxies, otherwise, negative AGN feedback should lead to statistically weaker narrow Hα emissions in Type 2 AGN.Based on the results in Figure 3, there are weaker narrow Hα emissions (larger values of R CL ) in the Type 2 AGN than in the starforming galaxies.Therefore, the results in Figure 3 can be accepted as apparent clues to support negative AGN feedback in the local Type 2 AGN in SDSS, considering the continuum emissions with few contaminations of central AGN activity in host galaxies of Type 2 AGN.
In order to confirm the shown results in Figure 3 leading to negative AGN feedback, the following effects are mainly considered.
If the lower line intensities of the intrinsic narrow Hα in Type 2 AGN were not due to the negative AGN feedback but due to central AGN activity, it would be necessary to check whether stronger AGN activity can lead to lower line intensities of intrinsic narrow Hα.If lower line intensities of intrinsic narrow Hα in Type 2 AGN were actually due to stronger AGN activity, statistically lower L Hα for the given λL cont could be expected in Type 2 AGN.However, in the collected Type 2 AGN, after checking the dependence of narrow Hα luminosity on central AGN activity traced by the narrow emission line ratio of O3HB, one positive correlation can be found with the Spearman's rank correlation coefficient of 0.31 (P null < 10 −15 ), shown in the right panel of Figure 2.Meanwhile, in the collected star-forming galaxies, there is a weak negative dependence of L Hα on the parameter of O3HB, with the Spearman's rank correlation coefficient of −0.22 (P null < 10 −15 ), also shown in the right panel of Figure 2. Therefore, the observed lower L Hα in the Type 2 AGN is not due to stronger central AGN activity.) in the star-forming galaxies (filled with red lines) and in the Type 2 AGN (filled with blue lines).In the left panel, solid and dashed blue lines, and solid and dashed red lines show the best-fitting results (the corresponding formula marked in the top-left corner) and the corresponding 1 RMS scatters to the correlations in the Type 2 AGN and the star-forming galaxies, respectively.
Furthermore, the effects of the different redshift distributions are considered in the results shown in Figure 3 between the star-forming galaxies and the Type 2 AGN.The left panel of Figure 5 shows the redshift distributions of the 19,351 starforming galaxies with a median z of about 0.047 ± 0.029 and the 3836 Type 2 AGN with a median z of about 0.065 ± 0.030.Uncertainties of the median values are the standard deviations of z.The student's t-test technique can be applied to confirm that the median values of z are different with a significance level higher than 5σ.Besides the different median redshifts, through the two-sided Kolmogorov-Smirnov statistic technique, the star-forming galaxies and the Type 2 AGN have the same redshift distributions with a significance level smaller than 10 −15 .Therefore, it is necessary to check the effects of the different redshift distributions on the results shown in Figure 3.In order to ignore the effects of the different redshift distributions, the simplest method is to check the results shown in Figure 3 but for two samples of star-forming galaxies and Type 2 AGN, which have the same redshift distributions.
Based on the redshift distributions of the star-forming galaxies and the Type 2 AGN, 7672 star-forming galaxies are easily and randomly collected into one subsample, which has a similar redshift distribution to that of the sample of the 3836 Type 2 AGN, as the results show in the right panel of Figure 5. Through the two-sided Kolmogorov-Smirnov statistic technique, the subsample of the 7672 star-forming galaxies and the 3836 Type 2 AGN have the same redshift distributions with a significance level higher than 99.99%.Then, the correlation between L Hα and λL cont is shown in the left panel of Figure 6 for the subsample of the 7672 star-forming galaxies.The linear correlation can be confirmed with the Spearman rank correlation coefficient of about 0.88 (P null < 10 −15 ) for the re-collected 7672 star-forming galaxies in the subsample.The LTS technique determined best-fitting results are shown in the left panel of Figure 6 with the corresponding formula marked in the top region of the left panel of Figure 6.It is clear that the lower expected L Hα for a given λL cont can be reconfirmed in the Type 2 AGN rather than in the star-forming galaxies in the subsample, as shown in the right panel of Figure 6 with the median values and the corresponding standard deviations of R log CL of about 2.40 ± 0.28 and 2.99 ± 0.34 in the 7672 starforming galaxies in the subsample and in the 3836 Type 2 AGN, respectively.The student's t-test technique can be applied to confirm the different median values of R log CL with a significance level higher than 5σ.Therefore, the very different correlations between L Hα and λL cont are intrinsic and reliable  between the star-forming galaxies and the Type 2 AGN, after considering the effects of different z distributions of the starforming galaxies and the Type 2 AGN.
Furthermore, combined with aperture effects, the effects of the different distributions of total stellar mass M T and different host galaxy morphologies are considered in the results shown in Figure 3 between the star-forming galaxies and the Type 2 AGN.The left panel of Figure 7 shows the M T distributions of the 18,995 star-forming galaxies (356 star-forming galaxies not included, due to their total stellar masses being smaller than five times their corresponding uncertainties) with a median M M log T ( )  of about 10.26 ± 0.63 and of the 3827 Type 2 AGN (36 Type 2 AGN not included, due to their total stellar masses smaller than five times their corresponding uncertainties) with a median M M log T ( )  of about 10.94 ± 0.41.Uncertainties of the median values are the standard deviations of M M log T ( )  .The student's t-test technique can be applied to confirm that the median values of M log T ( ) are different with a significance level higher than 5σ.Through the two-sided Kolmogorov-Smirnov statistic technique, the star-forming galaxies and the Type 2 AGN have the same redshift distributions with a significance level smaller than 10 −15 .Therefore, it is necessary to check the effects of the different M T distributions on the results shown in Figure 3.
Meanwhile, the middle panel of Figure 7 shows the R 90 (as the radii containing 90% of the Petrosian flux in the SDSS r band5 ) distributions of the 18,131 star-forming galaxies (1219 star-forming galaxies not included, due to their petroR90_r smaller than five times their corresponding uncertainties) with a median R log arcsec ( ) are different with a significance level higher than 5σ.Through the two-sided Kolmogorov-Smirnov statistic technique, the star-forming galaxies and the Type 2 AGN have the same R 90 distributions with a significance level smaller than 10 −15 .Therefore, it is necessary to check the effects of the different R 90 distributions on the results shown in Figure 3.
Meanwhile, the right panel of Figure 7 shows the distributions of the IC parameter (Shimasaku et al. 2001;Strateva et al. 2001) IC = R 50 /R 90 (where R 50 and R 90 as the radii containing 50% and 90% of the Petrosian flux in the  ( ) distributions of the collected 18,995 star-forming galaxies (filled with red lines) and the collected 3827 Type 2 AGN (filled with blue lines).The middle and right panels show the distributions of the R 90 and IC of the 18,131 star-forming galaxies and the 3652 Type 2 AGN with reliable measurements of parameters of petroR90_r and petroR50_r.
SDSS r band6 ) of the 18,131 star-forming galaxies (with reliable measurements of petroR90_r and petroR50_r) and the 3652 Type 2 AGN (with reliable measurements of petroR90_r and petroR50_r).The median values of IC log( ) are about −0.39 ± 0.05 and −0.45 ± 0.06 for the star-forming galaxies and the Type 2 AGN, respectively.The student's t-test technique can be applied to confirm that the median values of IC log( ) are different with a significance level higher than 5σ.Through the two-sided Kolmogorov-Smirnov statistic technique, the star-forming galaxies and the Type 2 AGN have the same IC distributions with a significance level smaller than 10 −15 .Therefore, it is necessary to check the effects of the different IC distributions on the results shown in Figure 3.
In order to check the effects of different distributions of z, M T , R 90, and IC, a simple method can be considered as follows by re-collected star-forming galaxies and the Type 2 AGN into new subsamples that have the same distributions of z, M T , R 90, and IC.The same distributions of z and R 90 can be applied to ignore aperture effects on the results for the objects in the subsamples.The same distribution of M T can be applied to ignore the effects of different total stellar masses on the results for the objects in the subsamples.And the same distributions of IC < 0.35 (combined with devab_r larger than 0.8) can be applied to simply accept that the host galaxies of the recollected objects are elliptical galaxies as discussed in Shimasaku et al. (2001), Strateva et al. (2001), and on the SDSS website7 in order to ignore effects of inclinations of disk galaxies and/or to ignore probable effects of different morphologies.Based on the z distributions (shown in the left panel of Figure 5) and the distributions of M T , R 90, and IC shown in Figure 7 of the star-forming galaxies and the Type 2 AGN, a subsample of 191 star-forming galaxies and a subsample of 191 Type 2 AGN can be collected, with the two subsamples having the same distributions of z, M T , R 90, and IC, which are shown in Figure 8. Through the two-sided Kolmogorov-Smirnov statistic technique, the subsample of the 191 star-forming galaxies and the subsample of the 191 Type 2 AGN have the same distributions of z, M T , R 90, and IC, with significance levels higher than 99.5%.
Then, the correlations between L Hα and λL cont are shown in the left panel of Figure 9 for the 191 star-forming galaxies and the 191 Type 2 AGN in the subsamples that have the same distributions of z, M T , R 90, and IC.The mean ratio of emission line flux to its corresponding uncertainty is about 36 in the narrow Hα, and the mean ratio of continuum emission intensity to its corresponding uncertainty is about 21.The linear correlations can be confirmed with the Spearman rank correlation coefficients of about 0.83 (P null < 10 −15 ) for the re-collected 191 star-forming galaxies in the subsample and of about 0.78 (P null < 10 −15 ) for the re-collected 191 Type 2 AGN in the subsample.The LTS technique determined best-fitting results are shown in the left panel of Figure 9 with the corresponding formula marked in the top region of the left panel of Figure 9.It is clear that the lower expected L Hα for the given λL cont can be reconfirmed in Type 2 AGN rather than in the star-forming galaxies in the subsamples, as shown in the right panel of Figure 9 with the median values and the corresponding standard deviations of R log CL of about 2.61 ± 0.24 and 2.98 ± 0.36 in the 191 star-forming galaxies and the 191 Type 2 AGN in the subsamples, respectively.The student's t-test technique can be applied to confirm the different median values of R log CL with significance levels higher than 5σ.Therefore, even after considering the effects of different distributions of z, M T , R 90, and IC, the very different correlations between L Hα and λL cont are intrinsic and reliable between the star-forming galaxies and the Type 2 AGN.
Moreover, due to unconfirmed positive or negative AGN feedback on star-formation histories, the following extreme comparisons can be checked between star-forming galaxies with lower (higher) total stellar masses and Type 2 AGN with higher (lower) total stellar masses.Among the 18,995 starforming galaxies with a median M M log T ( )  of about 10.26, there are half of the star-forming galaxies in the SF_H sample with total stellar masses M M log T ( )  larger than the median value of 10.26, and the other half of the star-forming galaxies in the SF_L sample with total stellar masses M M log T ( ) ( )  smaller than the median value 10.94.Then, the left panel of Figure 10 shows the distributions of R log CL of the 9478 star-forming galaxies in the SF_H sample with a median R log CL of about 2.53 ± 0.24 and the 1918 Type 2 AGN in the AGN_L sample with a median R log CL of about 2.88 ± 0.35; the right panel of Figure 10 shows the distributions of the R log CL of the 9477 star-forming galaxies in the SF_L sample with a median R log CL of about 2.25 ± 0.26 and the 1918 Type 2 AGN in the AGN_L sample with a median R log CL of about 3.07 ± 0.30.The student's t-test technique can be applied to confirm the different median values of R log CL shown in Figure 10 with significance levels higher than 5σ.Therefore, the very different correlations between L Hα and λL cont are intrinsic and reliable between star-forming galaxies and Type 2 AGN, even after considering very different M T distributions of the star-forming galaxies and the Type 2 AGN.
Moreover, the parameters of Dn(4000) and Hδ A (Worthey & Ottaviani 1997;Balogh et al. 1999) are compared between the 206 star-forming galaxies and the 206 Type 2 AGN in the subsamples with the same distributions of z, M T , R 90 , and IC (IC < 0.35 and devab_r larger than 0.8) because the two parameters can provide intrinsic information on stellar ages (Kauffmann et al. 2003).Here, the parameters Dn(4000) and Hδ A are measured through the spectroscopic features similar to what has been done in Kauffmann et al. (2003).The results on Dn(4000) and Hδ A are shown in Figure 11.Median values and the corresponding standard deviations of Dn(4000) and Hδ A are 1.57± 0.17 and 0.24 ± 2.02 in the 206 Type 2 AGN in the subsample, and 1.41 ± 0.13 and 1.99 ± 1.45 in the 206 starforming galaxies in the subsample, indicating older stellar populations in the host galaxies in the Type 2 AGN.The student's t-test technique can be applied to confirm that the mean values of Dn(4000) (Hδ A ) are different with significance levels higher than 5σ between the star-forming galaxies and the Type 2 AGN in the subsamples with the same distributions of z, M T , R 90 , and IC (IC < 0.35 and devab_r larger than 0.8).If there were positive AGN feedback on star formation in the host galaxies of Type 2 AGN, younger stellar ages should clearly be expected.Therefore, the results shown in Figure 11 can be accepted as indirect evidence to support the globally negative AGN feedback on star formation.
Furthermore, as discussed in Cid Fernandes et al. (2011), Type 2 AGN and star-forming galaxies can be separated in the space of EW(Hα) (equivalent width of narrow Hα) and N2HA; therefore, it is necessary to check whether the different R CL are only due to very different distributions of EW(Hα) between star-forming galaxies and Type 2 AGN.The top-left panel of Figure 12 shows the log EW Ha ( ( )) distributions, with a mean value of about 1.45 ± 0.32 for the star-forming galaxies and   4000) are about 10.78 ± 0.41, -0.44 ± 0.05, and 1.49 ± 0.16 for the 2013 Type 2 AGN in the subsample, and about 10.56 ± 0.54, -0.40 ± 0.06 and 1.41 ± 0.13 for the 2013 star-forming galaxies in the subsample.The student's t-test technique can be applied to confirm that the mean values of host galaxy properties are different enough with significance levels higher than 5σ.Through the two-sided Kolmogorov-Smirnov statistic technique, the significance levels were smaller than 10 −15 for the objects in the subsamples with the same distributions of M T , IC, and Dn(4000).In other words, besides the different EW(Hα) distributions, AGN feedback could play a key role leading to the different R CL .
Meanwhile, as shown above, on the steeper dependence of L Hα and λL cont in Type 2 AGN, it is necessary to check whether the more luminous Type 2 AGN included in our final sample can affect the statistical results of R CL .The left panel of Figure 13 shows the L log ergs ) with significance levels higher than 99.99%.Then, the right panel of Figure 13 shows the distributions of R log CL ( ), with a mean value of about 2.42 ± 0.28 for the star-forming galaxies and about 2.83 ± 0.29 for the Type 2 AGN in the subsamples.The student's t-test technique can be applied to confirm that the mean values of R log CL ( ) are different enough with significance levels higher than 5σ.In other words, the more luminous Type 2 AGN cannot be applied to explain the apparent different R log CL ( ) between the star-forming galaxies and the Type 2 AGN.
Certainly, in this manuscript, it has been accepted that there is no contribution of central AGN emission to the continuum emission of host galaxies of Type 2 AGN.If considering the contribution of AGN to the continuum emission leading to the different correlations between L Hα and λL cont in star-forming galaxies and Type 2 AGN, about 58% of continuum emission from central AGN activity could be expected in Type 2 AGN.So a larger contribution should indicate stronger power-law continuum emission components in the SDSS spectra of Type 2 AGN.However, after checking the SDSS spectra of the Type 2 AGN, there are no broad emission lines nor power-law  continuum components detected in the SDSS spectra of the Type 2 AGN.The flux-weighted mean spectrum of the Type 2 AGN is shown in Figure 14, which can be described by pure stellar templates without the consideration of power-law components, and have no apparent broad emission lines, to simply support that only a small number of Type 2 AGN have broad emission lines and AGN continuum emissions included in the SDSS spectra.Furthermore, if there was a strong contribution of central AGN activity to the continuum emission in the host galaxies of Type 2 AGN, the calculated Hδ A could be about two times different from the results shown in the left panel of Figure 11 for the Type 2 AGN, leading to unexpected quite different correlations between Dn(4000) and Hδ A in the star-forming galaxies and the host galaxies of Type 2 AGN.

Summary and Conclusions
The final summary and conclusions are as follows.
1.Among all the low-redshift narrow emission line main galaxies in SDSS DR16 with S/N > 20, the spectroscopic narrow emission lines of Hβ, [O III], Hα, and star-forming galaxies, a strong linear correlation can be confirmed between the narrow Hα line luminosity and the continuum luminosity at 5100 Å, with no contribution from AGN activity.The correlation in star-forming galaxies can be treated as a standard candle.4. Based on the reliably measured parameters of the 4112 Type 2 AGN, a strong correlation can be confirmed between the narrow Hα line luminosity and the continuum luminosity at 5100 Å, however, with quite a different intercept from the correlation in the star-forming galaxies.5. Star-forming galaxies (Type 2 AGN) with S/N > 30 can lead to a similar correlation between narrow Hα line The bottom panels show the distributions of M T , IC, and Dn(4000) of the star-forming galaxies and the Type 2 AGN in the subsamples that have the same distributions of z and EW(Hα).In each panel, a histogram filled with red lines shows the results for the star-forming galaxies, and a histogram filled with blue lines shows the results for the Type 2 AGN.
Figure 13.The left panel shows the distributions of λL cont of the star-forming galaxies and the Type 2 AGN.The second panel and the third panel show the distributions of redshift and λL cont of the star-forming galaxies and the Type 2 AGN in the subsamples, which have the same distributions of z and λL cont .The right panel shows the R log CL ( ) distributions of the star-forming galaxies and the Type 2 AGN in the subsamples, which have the same distributions of z and λL cont .In each panel, a histogram filled with red lines shows the results for the star-forming galaxies, and a histogram filled with blue lines represents the Type 2 AGN.luminosity and continuum luminosity as those of starforming galaxies (Type 2 AGN) with S/N > 30, leading to few effects of spectral S/N on the final results.6. Statistically lower narrow Hα line luminosities with a significance level higher than 5σ can be confirmed for the given continuum luminosities in Type 2 AGN than in the star-forming galaxies leading to apparent clues to support globally negative AGN feedback in Type 2 AGN, even after considering the effects of central AGN activity on the narrow Hα line luminosities, and the effects of different distributions of redshift z, total stellar mass M T , Petrosian radius R 90 , and inverse concentration parameter IC. 7. Different host galaxy properties can be confirmed with significance levels higher than 5σ between the starforming galaxies and the Type 2 AGN in the subsamples that have the same distributions of redshift and EW(Hα).Therefore, besides the different EW(Hα) properties, AGN feedback could play a key role in leading to the statistically lower narrow Hα line luminosities for the given continuum luminosities in the Type 2 AGN than in the star-forming galaxies.8. Comparing the properties of Dn(4000) and IC parameter between the 206 star-forming galaxies and the 206 Type 2 AGN in the final subsamples with the same distributions of z, M T , R 90, and IC (IC < 0.35 and devab_r larger than 0.8), statistically older stellar ages in Type 2 AGN can provide indirect evidence to support negative AGN feedback in the local Type 2 AGN. 9.Under the extreme assumption of no contribution of central AGN activity in the observed narrow Hα in the Type 2 AGN, the globally lower limited value can be estimated in which more than 50% of narrow Hα emissions have been suppressed in the host galaxies of the local SDSS Type 2 AGN due to globally negative AGN feedback.

Figure 2 .
Figure2.The left panel shows the BPT diagram of O3HB vs. N2HA for all the narrow emission line main galaxies collected from the SDSS DR16, with the collected 4112 Type 2 AGN lying above the dividing line shown as a solid dark green line and the collected 19,351 star-forming galaxies lying below the dividing line shown as the solid purple line.The right panel shows the dependence of L Hα on the parameter O3HB.In the right panel, the contour filled with different shades of green represents the results in the Type 2 AGN, and the contour with contour levels in reddish lines represents the results in the star-forming galaxies.In each panel, the color bar is shown in the top region to represent the corresponding number densities related to different colors.

Figure 1 .
Figure 1.The left panels show examples of the SSP method determined by stellar continuum (in the top-left panel) and the corresponding line spectrum (in the bottom-left panel) in the Type 2 AGN SDSS 0533-51994-0031.The right panels show the corresponding best-fitting results to the emission lines, after the subtraction of the determined stellar continuum.In the left panels, solid lines in dark green, red, and blue represent the observed SDSS spectrum, the SSP method determined the stellar continuum and the line spectrum after the subtraction of the stellar continuum, respectively.In the right panels, the solid dark green line shows the line spectrum, and the solid red line shows the best-fitting results.In the top-right panel, the solid purple line shows the determined narrow Hβ, the solid cyan lines, and solid blue lines show the determined core and shifted-wing related broad components of the [O III] doublet, respectively.In the bottom-right panel, the solid purple and cyan lines show the determined narrow Hα and [N II] doublet, respectively.In the top-left panel, in order to show clearer absorption features, the Y-axis is shown in logarithmic coordinates.

Figure 3 .
Figure3.The left panel shows the correlation between L Hα and λL cont in the 19,351 star-forming galaxies (the contour with contour levels in reddish colors) and in the 3836 Type 2 AGN (the contour filled with different shades of green).The right panel shows the distributions of R log CL ( ) in the star-forming galaxies (filled with red lines) and in the Type 2 AGN (filled with blue lines).In the left panel, solid and dashed blue lines, and solid and dashed red lines show the best-fitting results (the corresponding formula marked in the top-left corner) and the corresponding 1 RMS scatters to the correlations in the Type 2 AGN and the star-forming galaxies, respectively.

Figure 4 .
Figure 4. Same as Figure3, but for the star-forming galaxies with S/N > 30 and the Type 2 AGN with S/N > 30.

Figure 5 .
Figure5.The left panel shows the redshift distributions of the collected 19,351 star-forming galaxies (filled with red lines) and the collected 3836 Type 2 AGN (filled with blue lines).The right panel shows the redshift distributions of the randomly re-collected 7672 star-forming galaxies (filled with red lines) in the subsample and the 3836 Type 2 AGN (filled with blue lines).

Figure 6 .
Figure 6.Similar to Figure3, but the star-forming galaxies are the re-collected 7672 star-forming galaxies in the subsample.

Figure 7 .
Figure 7.The left panel shows the M log T( ) distributions of the collected 18,995 star-forming galaxies (filled with red lines) and the collected 3827 Type 2 AGN (filled with blue lines).The middle and right panels show the distributions of the R 90 and IC of the 18,131 star-forming galaxies and the 3652 Type 2 AGN with reliable measurements of parameters of petroR90_r and petroR50_r.

Figure 8 .
Figure 8. Distributions of the M T (top-left panel), z (top-right panel), R 90 (bottom-left panel), and IC (bottom-right panel) of the 191 star-forming galaxies and the 191 Type 2 AGN in the subsamples with the same distributions of M T , z, R 90 , and IC (IC < 0.35 and devab_r larger than 0.8).In each panel, a histogram filled with red lines shows the results for the star-forming galaxies, and a histogram filled with blue lines shows the results for the Type 2 AGN.

Figure 9 .
Figure 9. Similar to Figure 3 and Figure6, but for the re-collected 191 star-forming galaxies and the re-collected 191 Type 2 AGN in the subsamples, which have the same distributions of M T , z, 90 , and IC (IC < 0.35 and devab_r larger than 0.8).In the left panel, the sold circles plus error bars in blue show the results for the 191 Type 2 AGN, and circles plus error bars in dark green show the results for the 191 star-forming galaxies.
with a mean value of about 43.15 ± 0.59 for the star-forming galaxies and about 43.45 ± 042 for the Type 2 AGN.Then, based on the L log cont l ( )and redshift distributions, one subsample including 2967 star-forming galaxies and one subsample including 2967 Type 2 AGN are created, with the same distributions of z and L log cont l ( )as shown in the second panel and the third panel of Figure 13.Through the two-sided Kolmogorov-Smirnov statistic technique, the two subsamples have the same distributions of z and L log cont l (

Figure 10 .
Figure 10.The left panel shows the distributions of the R log CL of the star-forming galaxies in the SF_H sample and Type 2 AGN in the AGN_L sample.The right panel shows the distributions of R log CL of the star-forming galaxies in the SF_L sample and the Type 2 AGN in the AGN_H sample.In each panel, a histogram filled with dark green lines shows the results for the star-forming galaxies, and a histogram filled with blue lines shows the results for the Type 2 AGN.

Figure 11 .
Figure 11.The properties of Dn(4000) and Hδ A in the 191 star-forming galaxies (open circles plus error bars in red) and the 191 Type 2 AGN (open circles plus error bars in blue) in the subsamples with the same distributions of z, M T , R 90 and IC (IC < 0.35 and devab_r larger than 0.8).Solid circles plus error bars in red and blue show the mean positions and the corresponding uncertainties (the standard deviation) of the star-forming galaxies and the Type 2 AGN, respectively.

Figure 12 .
Figure12.The top-left panel shows the distributions of EW(Hα) of the star-forming galaxies and the Type 2 AGN.The top middle panel and the top-right panel show the distributions of the redshift and EW(Hα) of the star-forming galaxies and the Type 2 AGN in the subsamples that have the same distributions of z and EW(Hα).The bottom panels show the distributions of M T , IC, and Dn(4000) of the star-forming galaxies and the Type 2 AGN in the subsamples that have the same distributions of z and EW(Hα).In each panel, a histogram filled with red lines shows the results for the star-forming galaxies, and a histogram filled with blue lines shows the results for the Type 2 AGN. .
[N II] are well measured, after the subtraction of the contribution of the host galaxy determined by the SSP method applied with 39 stellar templates.2. Based on the dividing lines applied in the BPT diagram of O3HB versus N2HA, 19,531 star-forming galaxies and 4112 Type 2 AGN are collected with reliable narrow emission lines.3. Based on the reliably measured parameters of the 19,531