On the stellar clustering and architecture of planetary systems
V. Adibekyan, N. C. Santos, O. D. S. Demangeon, J. P. Faria, S. C. C. Barros, M. Oshagh, P. Figueira, E. Delgado Mena, S. G. Sousa, G. Israelian, T. Campante, A. A. Hakobyan
AAstronomy & Astrophysics manuscript no. adibekyan_2020_planet_environment_v4 © ESO 2021February 25, 2021
On the stellar clustering and architecture of planetary systems
V. Adibekyan , , N. C. Santos , , O. D. S. Demangeon , , J. P. Faria , , S. C. C. Barros , , M. Oshagh , ,P. Figueira , , E. Delgado Mena , S. G. Sousa , G. Israelian , , T. Campante , , and A. A. Hakobyan Instituto de Astrofísica e Ciências do Espaço, Universidade do Porto, CAUP, Rua das Estrelas, 4150-762 Porto, Portugale-mail: [email protected] Departamento de Física e Astronomia, Faculdade de Ciências, Universidade do Porto, Rua do Campo Alegre, 4169-007 Porto,Portugal European Southern Observatory, Alonso de Córdova 3107, Vitacura, Región Metropolitana, Chile Instituto de Astrofísica de Canarias, E-38205 La Laguna, Tenerife, Spain Departamento de Astrofìsica, Universidad de La Laguna, E-38206 La Laguna, Tenerife, Spain Center for Cosmology and Astrophysics, Alikhanian National Science Laboratory, 2 Alikhanian Brothers Str., 0036 Yerevan, Ar-meniaReceived date / Accepted date
ABSTRACT
Context.
Revealing the mechanisms shaping the architecture of planetary systems is crucial for our understanding of their formationand evolution. In this context, it has been recently proposed that stellar clustering might be the key in shaping the orbital architectureof exoplanets.
Aims.
The main goal of this work is to explore the factors that shape the orbits of planets.
Methods.
We perform di ff erent statistical tests to compare the properties of planets and their host stars associated with di ff erent stellarenvironments. Results.
We used a homogeneous sample of relatively young FGK dwarf stars with RV detected planets and tested the hypothesis thattheir association to phase space (position-velocity) over-densities (‘cluster’ stars) and under-densities (‘field’ stars) impacts the orbitalperiods of planets. When controlling for the host star properties, on a sample of 52 planets orbiting around ‘cluster’ stars and 15 planetsorbiting around ’field’ star, we found no significant di ff erence in the period distribution of planets orbiting these two populationsof stars. By considering an extended sample of 73 planets orbiting around ’cluster’ stars and 25 planets orbiting ’field’ stars, asignificant di ff erent in the planetary period distributions emerged. However, the hosts associated to stellar under-densities appeared tobe significantly older than their ’cluster’ counterparts. This did not allow us to conclude whether the planetary architecture is relatedto age, environment, or both. We further studied a sample of planets orbiting ‘cluster’ stars to study the mechanism responsible forthe shaping of orbits of planets in similar environments. We could not identify a parameter that can unambiguously be responsible forthe orbital architecture of massive planets, perhaps, indicating the complexity of the issue. Conclusions.
Increased number of planets in clusters and in over-density environments will help to build large and unbiased sampleswhich will then allow to better understand the dominant processes shaping the orbits of planets.
Key words.
Methods: statistical – Planets and satellites: formation – Planet-star interactions – Stars: fundamental parameters
1. Introduction
Understanding the mechanisms shaping the architecture of plan-etary systems is crucial to complete the picture of planet forma-tion and evolution (e.g. Winn & Fabrycky 2015; Hatzes 2016).Among the many open questions in this field, it is of particularinterest to understand the origin of hot Jupiters (HJs, Dawson &Johnson 2018) - short period giant planets . Several mechanismsare proposed to explain the presence of these massive planets atvery close distances to their host stars: in-situ formation, diskmigration, high-eccentricity tidal migration, and dynamical per-turbations by stellar fly-bys in open clusters. Although, a com-bination of these mechanisms might be needed to explain theobservational properties of HJs and their hosts stars (Dawson &Johnson 2018), it was very recently suggested that the short pe- The definition of HJs in terms of upper limit of orbital period (orsemi-major axis) and lower limit in planetary mass (or radius) varies inthe literature. riods of HJs originate from environmental perturbations (Winteret al. 2020, hereafter, W20 ).To study the possible link between stellar clustering and ar-chitecture of planetary systems, Winter et al. (2020) estimatedthe probability ( P high ) that a planet host star belongs to over-or under-densities in the position-velocity phase space. The au-thors determined and made publicly available the P high values formore than 1500 exoplanet host stars for which radial velocitieswere available in Gaia DR2. Stars with P high > .
84 were con-sidered as potential members of co-moving groups (over-densityor ‘cluster’ stars) and stars with P high < .
16 as ‘field’ stars.Based on this database, they reached to two important conclu-sions: Planets orbiting stars associated with over-densities havesignificantly shorter orbital periods than those orbiting around‘field’ stars and that HJs predominantly exist around ‘cluster’stars.Given the importance of these findings and conclusions, inthis manuscript we performed an independent analysis of their
Article number, page 1 of 10 a r X i v : . [ a s t r o - ph . E P ] F e b & A proofs: manuscript no. adibekyan_2020_planet_environment_v4 data but using homogeneously determined stellar parameters ofthe planet host stars from the SWEET-Cat (Santos et al. 2013).The manuscript is organized as follows: In Sect. 2 we firstbuilt a homogeneous sample trying to control di ff erent biasesand then studied the period distribution of planets orbiting ‘clus-ter’ and ‘field’ stars. In Sect. 3 we studied the impact of di ff erentphysical parameters on the orbital periods of planets associatedwith high-density stellar environments. We summarize our workin Sect. 4.
2. SWEET-Cat FGK dwarf RV sample
In order to confirm or refute the main findings of Winter et al.(2020) it is crucial to perform the analysis on an unbiased sam-ple. For a discussion about the impact of di ff erent (potential) bi-ases we refer the reader to W20. In this section we build a sam-ple (based on the original full sample of W20) of RV detected high-mass planets orbiting around FGK dwarf stars for whichhomogeneously derived (see Sect. A) stellar parameters exist inSWEET-Cat. We then perform a statistical analysis on this datausing the AD test to study the impact of stellar clustering on theorbital periods of exoplanets. It is important to note that by re-stricting the sample to RV detected planets we did not removethe observational biases that this planet detection method su ff ersfrom. However, we try to minimize the impacts of di ff erent bi-ases by applying further restrictions on the properties of planetsand their hosts. Ideally, one would have to carefully model andcorrect for the detection biases to construct the actual period dis-tributions of the planets. However, this would be extremely dif-ficult since the planets of our sample come from di ff erent planetsearch programs carried out with di ff erent instruments, di ff erentobservational strategies, and di ff erent detection biases.In this analysis we focus only on massive planets withmasses between 50 M ⊕ and 4 M jup . The selected lower limit isthe same as the one adopted in W20 for HJs. This semi-arbitrarylimit (see the discussion in Adibekyan 2019) is considered todecrease the impact of planet detection limits (low-mass planetsare di ffi cult to detect in wide orbits) and also the planet core-accretion models predicted a minimum in the planetary mass-distribution at about 50 M ⊕ mass (e.g. Mordasini et al. 2009).Our choice of upper mass limit is motivated by the recent find-ings that the properties of stars hosting super-massive Jupiters( M pl > M jup ) are di ff erent from those hosting lower-massJupiters which might suggest a di ff erent formation mechanisms(e.g. Santos et al. 2017; Adibekyan 2019; Maldonado et al. 2019;Goda & Matsuo 2019). The range of e ff ective temperature of theselected FGK stars is 4500 < T e ff < g < . ff erentdistributions when compared with the properties of planets orbit-ing around dwarfs (e.g. Adibekyan et al. 2013; Maldonado et al.2013; Mortier et al. 2013).The aforementioned constraints lead us to a sample of 178FGK dwarf stars hosting 214 RV detected giant planets . Forall these stars we homogeneously determined the isochrone ages Transiting planets have only short periods and are not suitable for ouranalysis. Note that due to the constraint on the homogeneity of the stellar pa-rameters only 10 stars (i.e. 5%) have been excluded at this stage. using the PARAM v1.3 web interface . The detaials of the agedetermination and the results of their comparison with the agesused in W20 are presented in Sect. B. We then applied the finalcut on age as suggested in W20 (stars with ages between 1 and4.5 Gyr) to build our main sample, hereafter called FGK P low , high sample. This sample consists of 44 P high5 (52 planets) and 14 P low (15 planets) stars. The distribution of these planets on thePeriod-Mass diagram is shown in the left panel of Fig. 1. P low and P high stars In Fig. C.1 we compare the CDFs of di ff erent properties of plan-ets and their host stars associated with over- and under-densities.The figure and the corresponding P AD values suggest that theplanets orbiting these two groups of stars do not have signifi-cantly di ff erent distributions of the orbital periods. The figurealso shows that the host star show significantly di ff erent distri-butions of T e ff and ages, the P high stars being hotter and youngerthan their P low counterparts. In particular, only 3 out of the 15planets orbiting P low stars are younger than 3 Gyr. The numberof planets (age < P high stars is 32, whichmakes about 60% (32 out of 52) of the whole sample.Although in the aforementioned analysis the AD test doesnot rejects the null hypothesis that overall distributions of peri-ods of planets orbiting P high and P low stars come from the sameparent distribution, Fig. 1 visually suggest an overabundance ofshort period planets (periods shorter than about 10 to 30 days)around P high stars when compared to their P low counterparts.Thefraction of short period (period <
30 days ) planets orbiting P high stars is 23.1 + . − . % (12 out of 52). This number, being slightlylarger, however, statistically speaking is not di ff erent from theone for the P low sample,: 20.0 + . − . % (3 out of 15). The di ff er-ence remains not significant if one considers more commonlyused period limit of 10 days for HJs (e.g. Wang et al. 2015):17.3 + . − . % (9 out of 52) and 20.0 + . − . % (3 out of 15) for the HJsorbiting around P high and P low stars, respectively.Unfortunately, by applying the cut on age and selecting onlyRV detected planets, we significantly reduced the size of thesample, especially the number of P low stars. The reduced sam-ple size has a direct impact on the errors of the estimated HJsfractions and might be responsible for the insignificance of theaforementioned di ff erences. Below we try to expand the sam-ple by increasing the range in stellar ages and relaxing the P high threshold.In Fig. 2 we show fraction of HJs orbiting stars associatedto over- and under-densities as a function of the P high thresholdwhich is used to separate the stars into this two categories. Inthe figure we considered 10, 20, and 30 days as the upper limitfor the orbital periods of HJs. The figure shows that the maxi-mum di ff erence of the fraction of HJs orbiting around ‘cluster’and ‘field’ stars is observed at the P high threshold of 0.3. Thisdi ff erence is however, is not significant at even one- σ level. Thefigure also shows that, as for the main sample, the two groups ofstars have significantly di ff erent distribution of ages.In the fol-lowing tests we will adopt the 0.3 value as the P high threshold, http: // stev.oapd.inaf.it / cgi-bin / param The stars with P high > P high , and the stars with P high < P low . By adopting an upper limit of 0.2 AU for the semi-major axis of HJs,Winter et al. (2020) limited their sample to planets with orbital periodsshorter than about 30 days.Article number, page 2 of 10. Adibekyan et al.: On the stellar clustering and architecture of planetary systems
50 200 1000 M pl ( M ) P e r i o d ( d a y s ) P high > 0.84 P high < 0.16
50 200 1000 M pl ( M ) P e r i o d ( d a y s ) P high > 0.7 P high < 0.3 Fig. 1.
Distribution of the RV detected planets on the Period-Mass diagram orbiting around FGK dwarf stars with homogeneously derived stellarparameters in SWEET-Cat. The left panel is for the FGK P low , high sample and the right panel is for the planets in the expended sample. which allows both to increase the sample size and decrease thecontamination of the samples by excluding the stars with inter-mediate P high probabilities.To further increase the sample size, we reduced the lowerage limit from 1 Gyr to 0.5 Gyr. Although individual planets orplanetary systems can show instabilities at timescales of a fewGyrs (in fact, one of the phenomena responsible for the instabil-ity on long timescales is the fly-by encounters that can occur indense stellar environments, Davies et al. 2014), usually the or-bits of massive planets become stable at less than about 100 Myr(e.g. Raymond et al. 2009; Davies et al. 2014; Sotiriadis et al.2017; Bitsch et al. 2020). As discussed in W20 (also see Krui-jssen et al. 2020), going beyond 5 Gyr leads to a strong contam-ination of ‘field’ sample by former over-density stars and shouldbe avoided. In general, the younger the stars the easier and morereliable its association to over- or under-density stellar environ-ments.Fig. 3 shows the fraction of HJs orbiting stars associated toover- and under-densities as a function of upper limit of stel-lar ages. The figure shows that the HJs fractions are statisticallyspeaking similar up to upper age limit of 5 Gyr. Moreover, thedi ff erence in the HJs fraction with age increases mostly becauseof the decrease of the fraction of HJs orbiting ‘field’ stars. This issomehow counter-intuitive, because as it was mentioned earlier,older ‘field’ samples are more contaminated by former ‘cluster’stars for which the fraction of HJs becomes higher. Up to agesof 3 Gyr, the fraction of HJs orbiting around ‘field’ stars is evenhigher than that for the ‘cluster’ stars. However, it is importantto note that the number of P low stars with ages below 3 Gyr isvery small. Fig. 3 also shows that when going beyond the 5 Gyrlimit, the HJ fraction slightly decreases for both P low and P high samples. This is because on average the hosts of HJs are slightlyyounger than the hosts of their longer period counterparts. Thisresult is similar to the one of Hamer & Schlaufman (2019) wherethe authors concluded that tidal interactions cause HJs to inspi-ral on a timescale shorter than the main sequence lifetime of thestars. Considering stars with ages between 0.5 and 5 Gyr, and the P high threshold of 0.3 we construct an extended sample consist-ing of 73 planets orbiting P high stars and 25 planets orbiting P low stars. The distributions of these planets in the Period-Mass dia-gram is shown in the right panel of Fig. 1. The di ff erence in theHJ fractions between the two groups is largest when consideringan upper orbital period limit of 30 days for HJs. This di ff erence(28.8 + . − . % (21 out of 73) and 12.0 + . − . % (3 out of 25) is signifi-cant at about 80% level, which would correspond to ∼ σ fora Gaussian distribution. However, it is important to stress againthe statistically significant di ff erence in age as inferred from thep-values of the AD test (see Fig. C.2).For the extended sample Fig. C.2 suggest a statistically sig-nificant di ff erence for the period distributions of planets orbitingstars in over- and under-densities. However, the two sub-samplesshow statistically di ff erent distributions in planetary mass, T e ff ,and stellar age. Restricting the sample (25 and 22 planets orbit-ing around P high and P low stars, respectively) to stars with agesbetween 2.5 and 5 Gyr and to planets with masses >
150 M ⊕ allows to vanish the di ff erences in planetary mass, T e ff , and stel-lar age. This restriction also dilutes the di ff erence in the orbitalperiod distributions (see Fig. C.3).
3. Properties of stars hosting short- versuslong-period planets in over- and under-densityenvironments
The analysis of the previous section did not reveal an unambigu-ous relation between stellar clustering and orbital architecture ofexoplanets. In this section, we separate the P high and P low sam-ples to study the impact of physical properties of the host starson the orbital properties of giant planets. In this way we elimi-nate the impact (if any) of stellar clustering on the architectureof planets.Adibekyan et al. (2013) showed that most of the massiveplanets orbiting low-metallicity stars ([Fe / H] < -0.1 dex) haveorbital periods longer than about 100 days (see also Sozzetti2004; Maldonado et al. 2012). The authors suggested that plan- Article number, page 3 of 10 & A proofs: manuscript no. adibekyan_2020_planet_environment_v4 P e r i o d < d a y s <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 P AD (Age) P e r i o d < d a y s P high threshold P e r i o d < d a y s F r a c t i o n o f H J s ( % ) Fig. 2.
Fraction of HJs orbiting ‘cluster’ (brown) and ‘field’ (skyblue)stars as a function of P high threshold which is used to separate the afore-mentioned two groups. The numbers next to the HJs fraction correspondto the numbers of HJs and the total number of planets that are used tocompute these fractions. The upper limit for the HJs orbital period is setat 10 (top), 20 (middle), and 30 (bottom) days. The P AD values on topof the panels show the results of AD test comparing the distributions ofthe ages of the ‘cluster’ and ‘field’ stars. The errorbars represent 68.3%(1- σ ) interval. ets in a metal-poor disk are forming further out and / or undergo-ing less migration as they take longer to form. Recently, Osborn& Bayliss (2020) studied the metallicity distribution of HJs andfound that although they preferentially orbit metal-rich stars, theaverage metallicity of their hosts is not higher than that of starshosting cold Jupiters. The authors concluded that hot and coldJupiters are formed in a similar process, but they have di ff erentmigration histories. In complement to these results, Dawson &Murray-Clay (2013) and Buchhave et al. (2018) showed that gi-ant planets orbiting metal-rich stars show signatures of dynami-cal interaction. Buchhave et al. (2018), in particular, showed thatHJs and cold eccentric Jupiters are preferentially orbiting metal-rich stars, while cold-Jupiters with circular orbits are mostly ob-served around solar-metallicity stars.On the left and right panels of Fig. 4 we separately showthe distribution of planets orbiting the P high and P low stars of theextended FGK P low , high sample. Due to the age constraints (youngstars are on average metallic), the number of planets orbitingmetal-poor stars (Fe / H] < P low stars is very small, we will nextfocus only on the planets orbiting P high stars. This sample con- P e r i o d < d a y s <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 P AD (Age) P e r i o d < d a y s Upper age limit P e r i o d < d a y s F r a c t i o n o f H J s ( % ) Fig. 3.
Fraction of HJs orbiting ‘cluster’ (brown) and ‘field’ (skyblue)stars as a function of upper age limit. The lower age limit is set to 0.5Gyr and the P high threshold is set to 0.3. The symbols and numbers arethe same as in Fig. 2. sists of 21 planets with periods shorter than 30 days and 52 plan-ets with longer periods. Fig. C.4 shows the CDFs of these short-and long-periods planets and their host stars. As indicated by the P AD values, the two groups are significantly di ff erent only in thedistribution of the planetary masses, the HJs having on averagelower-masses. If the upper period limit is reduced to 10 or 20days, the results remain practically the same. The fact that theage distribution of the short and long period planets are similarmight indicate that tidal inspiral is not significantly depleting theshort period planets in this age range. However, a dedicated anal-ysis on a larger sample is required to make a firm conclusion.Unfortunately, perhaps due to the small size of the sampleand / or due to the complexity of the problem, it is di ffi cult tofirmly conclude which parameter(s) internal for the star-planetsystem is(are) responsible for the period distribution of exoplan-ets orbiting stars formed in similar stellar environment.
4. Summary and conclusion
Very recently Winter et al. (2020) performed a tremendous workby assigning a large sample of exoplanet host stars to low-( P low ) or high-density ( P high ) stellar environments. The authorsthen used this sample to conclude that planets orbiting stars inhigh-density environments have significantly shorter periods andsmaller semi-major axis than their counterparts orbiting ‘field’(low-density) stars. They also found that most of the hot Jupitersare orbiting around ‘cluster’ stars. These findings, if confirmed,may have very important implications for our understanding ofplanet formation and evolution.In this manuscript we constructed a sample of FGK dwarfstars with only RV detected HJs for which homogeneously deter- Article number, page 4 of 10. Adibekyan et al.: On the stellar clustering and architecture of planetary systems
50 200 1000
M ( M ) P e r i o d ( d a y s ) P high stars [Fe/H] >= 0.0[Fe/H] < 0.0
50 200 1000
Mass ( M ) P low stars [Fe/H] >= 0.0[Fe/H] < 0.0 Fig. 4.
Period-Mass diagram of planets around FGK P high (left panel) and P low (right panel) dwarf stars with homogeneously derived stellarparameters in SWEET-Cat. Only stars with ages between 1 and 4.5 Gyr are shown. Planets orbiting metal-poor and metal-rich stars are shown inopen and filled circles, respectively. mined stellar parameters are available in the SWEET-Cat catalog(Santos et al. 2013). Additionally, we made a further constraintson the upper mass of planets at 4 M jup since the origin of thesuper-massive planets might be di ff erent (e.g. Santos et al. 2017;Adibekyan 2019). For this sample of stars we homogeneouslydetermined isochrone ages.In this small but significantly less biased sample of stars withages between 1 and 4.5 Gyr (52 planets orbiting P high stars and15 planets orbiting P low stars), we found no significant di ff er-ence in the period distribution of planets orbiting P high and P low stars. We then constructed an extended sample by slightly relax-ing the constrains on age and the P high threshold. In this sample,consisting of 73 planets orbiting around ’cluster’ stars and 25planets orbiting around ’filed’ stars, we found a statistically sig-nificant di ff erence for the period distributions of planets orbitingaround these two populations of stars. However, the ‘field’ and‘cluster’ stars also showed a significant di ff erence in the stellarage. When controlling for the host star properties, the di ff erencesin orbital periods of planets orbiting around stars associated withthe over- and under-densities diminishes. Thus, it is not possibleto conclude whether the planetary architecture is related to age,environment, or both.Next we focused only on a sub-sample of planets orbiting P high stars with the aim of understanding the mechanism respon-sible for shaping their planetary orbits in similar environments.We could not identify a parameter that unambiguously can beresponsible for the orbital architecture of these planets.It is important to note that although our analysis does notsuggest that the stellar clustering is the key parameter shapingthe orbits of planets, it still can play a role, especially given someobservational (e.g. Brucalassi et al. 2016) and theoretical (e.g.Shara et al. 2016; Wang et al. 2020) support of this hypothesis.The full picture of planet survival in dense stellar environmentsis not simple and depends on many external and internal to star-planets factors (e.g. Stock et al. 2020, and references therein). In-creased number of planet hosts in clusters and in over-density en-vironments will help to build large and unbiased samples whichwill then shed a light on this issue. Acknowledgements.
We thank the anonymous referee for the very con-structive comments and suggestions which helped us to substantiallyimprove the quality of the work amd presentations of the results.This work was supported by FCT - Fundação para a Ciência e Tec-nologia (FCT) through national funds and by FEDER through COM-PETE2020 - Programa Operacional Competitividade e Internacionalizaçãoby these grants: UID / FIS / / / / / / / FIS-AST / / / FIS-AST / / / / / CP1273 / CT0001, IF / / / CP1273 / CT0003,IF / / / CP0150 / CT0002, and IF / / / CP1215 / CT0002, respec-tively, and POPH / FSE (EC) by FEDER funding through the program “ProgramaOperacional de Factores de Competitividade - COMPETE”. O.D.S.D. and J.P.F.are supported in the form of work contracts (DL 57 / / CP1364 / CT0004 andDL57 / / CP1364 / CT0005, respectively) funded by FCT. T.C. is supportedby Fundação para a Ciência e a Tecnologia (FCT) in the form of a work con-tract (CEECIND / / References
Adibekyan, V. 2019, Geosciences, 9, 105Adibekyan, V. Z., Figueira, P., Santos, N. C., et al. 2013, A&A, 560, A51Astropy Collaboration, Price-Whelan, A. M., Sip˝ocz, B. M., et al. 2018, AJ, 156,123Bitsch, B., Trifonov, T., & Izidoro, A. 2020, A&A, 643, A66Bressan, A., Marigo, P., Girardi, L., et al. 2012, MNRAS, 427, 127Brucalassi, A., Pasquini, L., Saglia, R., et al. 2016, A&A, 592, L1Buchhave, L. A., Bitsch, B., Johansen, A., et al. 2018, ApJ, 856, 37da Silva, L., Girardi, L., Pasquini, L., et al. 2006, A&A, 458, 609Davies, M. B., Adams, F. C., Armitage, P., et al. 2014, in Protostars and PlanetsVI, ed. H. Beuther, R. S. Klessen, C. P. Dullemond, & T. Henning, 787Dawson, R. I. & Johnson, J. A. 2018, ARA&A, 56, 175Dawson, R. I. & Murray-Clay, R. A. 2013, ApJ, 767, L24Gaia Collaboration, Brown, A. G. A., Vallenari, A., et al. 2018, ArXiv e-prints[ arXiv:1804.09365 ]Goda, S. & Matsuo, T. 2019, ApJ, 876, 23Hamer, J. H. & Schlaufman, K. C. 2019, AJ, 158, 190Hatzes, A. P. 2016, Space Sci. Rev., 205, 267
Article number, page 5 of 10 & A proofs: manuscript no. adibekyan_2020_planet_environment_v4
Hunter, J. D. 2007, Computing in Science and Engineering, 9, 90Kruijssen, J. M. D., Longmore, S. N., & Chevance, M. 2020, ApJ, 905, L18Maldonado, J., Eiroa, C., Villaver, E., Montesinos, B., & Mora, A. 2012, A&A,541, A40Maldonado, J., Villaver, E., & Eiroa, C. 2013, A&A, 554, A84Maldonado, J., Villaver, E., Eiroa, C., & Micela, G. 2019, A&A, 624, A94Mordasini, C., Alibert, Y., Benz, W., & Naef, D. 2009, A&A, 501, 1161Mortier, A., Santos, N. C., Sousa, S. G., et al. 2013, A&A, 557, A70Osborn, A. & Bayliss, D. 2020, MNRAS, 491, 4481Pettitt, A. N. 1976, Biometrika, 63, 161Raymond, S. N., Armitage, P. J., & Gorelick, N. 2009, ApJ, 699, L88Saculinggan, M. & Balase, E. A. 2013, Journal of Physics: Conference Series,435, 012041Santos, N. C., Adibekyan, V., Figueira, P., et al. 2017, A&A, 603, A30Santos, N. C., Sousa, S. G., Mortier, A., et al. 2013, A&A, 556, A150Shara, M. M., Hurley, J. R., & Mardling, R. A. 2016, ApJ, 816, 59Sotiriadis, S., Libert, A.-S., Bitsch, B., & Crida, A. 2017, A&A, 598, A70Sousa, S. G., Adibekyan, V., Delgado-Mena, E., et al. 2018, A&A, 620, A58Sousa, S. G., Santos, N. C., Mayor, M., et al. 2008, A&A, 487, 373Sozzetti, A. 2004, MNRAS, 354, 1194Stock, K., Cai, M. X., Spurzem, R., Kouwenhoven, M. B. N., & Portegies Zwart,S. 2020, MNRAS, 497, 1807van der Walt, S., Colbert, S. C., & Varoquaux, G. 2011, Computing in Scienceand Engineering, 13, 22Virtanen, P., Gommers, R., Oliphant, T. E., et al. 2020, Nature Methods, 17, 261Wang, J., Fischer, D. A., Horch, E. P., & Huang, X. 2015, ApJ, 799, 229Wang, Y.-H., Leigh, N. W. C., Perna, R., & Shara, M. M. 2020, ApJ, 905, 136Wes McKinney. 2010, in Proceedings of the 9th Python in Science Conference,ed. Stéfan van der Walt & Jarrod Millman, 56 – 61Winn, J. N. & Fabrycky, D. C. 2015, ARA&A, 53, 409Winter, A. J., Kruijssen, J. M. D., Longmore, S. N., & Chevance, M. 2020, Na-ture, 586, 528
Article number, page 6 of 10. Adibekyan et al.: On the stellar clustering and architecture of planetary systems [Fe/H] SWEET-Cat [ F e / H ] ( S W EE T - C a t - N E A ) P high > 0.84 P high < 0.16 [Fe/H] SWEET-Cat C D F P KS = 0.0001 0.6 0.0 0.6 [Fe/H] NEA P KS = 0.0050.4 0.2 0.0 0.2 0.4 [Fe/H] SWEET-Cat [ F e / H ] ( S W EE T - C a t - N E A ) P high > 0.84 P high < 0.16 [Fe/H] SWEET-Cat C D F P KS = 0.07 0.4 0.0 0.4 [Fe/H] NEA P KS = 0.63 Fig. A.1.
Comparison of the stellar metallicities taken from the NASAExoplanet Archive (NEA) and homogeneously derived in SWET-Cat.The top and bottom panels show the results for the full and main sam-ples, respectively. The brown symbols and curves represent the data forthe P high stars and the skyblue symbols and curves for the P low stars.The right panels show the CDFs of the metallicities for the P low and P high stars. Appendix A: Importance of homogeneity of stellarparameters
When performing a statistical analysis of the properties of starswith and / or without planets, it is important to use parameters ashomogeneously derived as possible (e.g. Adibekyan 2019). Ex-oplanet archives and catalogs usually consist of heterogeneouscompilation of stellar properties which might lead to significantdiscrepancies when compared with homogeneously derived pa-rameters (e.g. Santos et al. 2013; Sousa et al. 2018). The hoststar properties listed in NEA (used by W20) are compiled fromdi ff erent sources. Moreover, while the physical parameters ofthe RV detected planets are mostly derived from high-resolutionspectra, such high-quality data do not necessarily exist for thetransiting planet hosts.We cross-matched the full (1421 planets orbiting around1058 stars) and the main (506 planets orbiting around 388 stars)samples of W20 with the SWEET-Cat catalog (Santos et al.2013; Sousa et al. 2018) which provides the stellar parametersof planet host stars. Although SWEET-Cat is one of the largestcatalog (and the largest one for the RV detected planets) of planethost stars with homogeneously determined stellar parameters,unfortunately it contains stellar parameters only for 375 starsfrom the full sample and 153 stars from the main sample ofW20. In Figs. A.1 and A.2 we compare the stellar metallicitiesand masses presented in NEA and homogeneously derived inSWEET-Cat catalog (Santos et al. 2013). The mean di ff erenceand dispersion for metallicity is 0.01 ± ± ff erence and dispersion is 0.0 ± M (cid:12) and0.0 ± M (cid:12) for the full and main samples, respectively.In the left panels of Figs. A.1 and A.2 for the P high and P low stars we compare the CDFs of metallicity and masses as takenfrom NEA and SWEET-Cat. We then performed a KS test toevaluate the similarities of the distributions. The P KS values for M star SWEET-Cat M s t a r ( S W EE T - C a t - N E A ) P high > 0.84 P high < 0.16 M star SWEET-Cat C D F P KS = 0.004 2 4 6 8 M star NEA P KS = 0.00021.0 1.5 2.0 2.5 M star SWEET-Cat M s t a r ( S W EE T - C a t - N E A ) P high > 0.84 P high < 0.16 M star SWEET-Cat C D F P KS = 0.22 1.0 1.5 2.0 M star NEA P KS = 0.15 Fig. A.2.
The same as Fig. A.1 but for stellar masses. most of the cases are very similar. The exception is for the stel-lar metallicity for the main sample, where there is a significantdi ff erence of P KS obtained for the NEA and SWEET-Cat val-ues. Furthermore, if instead of the KS test the Anderson-Darling(AD) test is performed to the aforementioned samples, the re-sult would di ff er more dramatically ( P AD = P AD > P high and P low stars do not come from the same par-ent distribution. Appendix B: Stellar ages
For the sample of 178 FGK dwarf stars hosting 214 RV de-tected giant planets (see Sect. 2) we derived the stellar ages fromthe PARAM v1.3 web interface based on the Padova theoreti-cal isochrones from Bressan et al. (2012) and with the use of aBayesian estimation method (da Silva et al. 2006). As input pa-rameters for PARAM, we used the Gaia DR2 parallaxes (GaiaCollaboration et al. 2018), V magnitudes extracted from Sim-bad , and spectroscopic T e ff and [Fe / H]. No correction for inter-stellar reddening was needed since all the stars are nearby ob-jects. The ages of all the stars is presented in a table at the CDS.In Fig. B.1 we compare the ages homogeniously derived inthis work and those from NEA. While practically there is no o ff -set (-0.1 Gyr) the dispersion is 2.7 Gyr. The figure also showsa group of 13 stars with NEA ages of exactly 1 Gyr. Eight ofthese stars, however, have isochrone ages (as derived in thiswork) greater than 5 Gyr and are excluded from the main sam-ple. All these stars all cool ( T e ff < g < . Appendix C: Supplementary figures AD is similar to the KS test, but is more sensitive to the tails ofdistribution and has higher power for small samples (Pettitt 1976; Sac-ulinggan & Balase 2013). http: // stev.oapd.inaf.it / cgi-bin / param http: // simbad.u-strasbg.fr / simbad / Article number, page 7 of 10 & A proofs: manuscript no. adibekyan_2020_planet_environment_v4
Age current work (Gyr) A g e N E A ( G y r ) Fig. B.1.
Comparison of the stellar ages taken from NEA and homo-geneously derived in this work. The skyblue dashed line represents theidentity line.Article number, page 8 of 10. Adibekyan et al.: On the stellar clustering and architecture of planetary systems
30 500 3000
Period (days) P AD =0.09 100 300 1000 M pl ( M ) P AD =0.27 1 2 3 4 5 Age (Gyr) P AD <0.01 0 50 100 150 Dist. (pc) P AD =0.460.3 0.0 0.3 [Fe/H] P AD =0.5 5000 5500 6000 T eff (K) P AD =0.04 4.2 4.4 4.6 4.8 log g P AD =0.46 0.8 1.0 1.2 Mass ( M ) P AD =0.5 C u m u l a t i v e d i t r i b u t i o n f un c t i o n Fig. C.1.
CDFs of di ff erent properties of planets and their host stars from the FGK P low , high sample associated with over- (brown) and under-densities(skyblue). The P AD values for each parameter is shown in the respective plot.
30 500 3000
Period (days) P AD =0.01 100 300 1000 M pl ( M ) P AD <0.01 1 2 3 4 5 Age (Gyr) P AD <0.01 0 50 100 150 Dist. (pc) P AD =0.110.3 0.0 0.3 [Fe/H] P AD =0.46 5000 5500 6000 T eff (K) P AD =0.01 4.2 4.4 4.6 4.8 log g P AD =1.00 0.8 1.0 1.2 Mass ( M ) P AD =0.32 C u m u l a t i v e d i t r i b u t i o n f un c t i o n Fig. C.2.
The sames as Fig. C.1 but for the extended sample: stars with ages between 0.5 and 5 Gyr, and the P high threshold of 0.30.Article number, page 9 of 10 & A proofs: manuscript no. adibekyan_2020_planet_environment_v4
30 500 3000
Period (days) P AD =0.14 100 300 1000 M pl ( M ) P AD =0.16 1 2 3 4 5 Age (Gyr) P AD =0.3 0 50 100 150 Dist. (pc) P AD =0.340.3 0.0 0.3 [Fe/H] P AD =0.94 5000 5500 6000 T eff (K) P AD =0.41 4.2 4.4 4.6 4.8 log g P AD =0.31 0.8 1.0 1.2 Mass ( M ) P AD =0.2 C u m u l a t i v e d i t r i b u t i o n f un c t i o n Fig. C.3.
The sames as Fig. C.1 but for a sample of stars with ages between 2.5 and 5 Gyr, planets with masses greater than 150 M ⊕ , and the P high threshold of 0.30.
10 40 200 2000
Period (days) P AD <0.01 100 300 1000 M pl ( M ) P AD <0.01 1 2 3 4 5 Age (Gyr) P AD = 0.66 0 50 100 150 Dist. (pc) P AD = 0.790.3 0.0 0.3 [Fe/H] P AD = 0.44 5000 5500 6000 T eff (K) P AD = 0.42 4.0 4.4 4.8 log g P AD = 0.42 0.75 1.00 1.25 Mass ( M ) P AD = 0.47 C u m u l a t i v e d i t r i b u t i o n f un c t i o n Fig. C.4.
CDFs of di ff erent properties of short- (Period <
30 days; solid line) and long-period (Period >