Can we predict if planets are lonely?

Title: The California-Kepler Survey. VI: Kepler Multis and Singles Have Similar Planet and Stellar Properties Indicating a Common Origin

Authors: Lauren M. Weiss, Howard T. Isaacson, Geoffrey W. Marcy, Andrew W. Howard, Erik A. Petigura, Benjamin J. Fulton, Joshua N. Winn, Lea Hirsch, Evan Sinukoff, Jason F. Rowe

First Author’s Institution: Université de Montréal, Canada

Status: Submitted to The Astronomical Journal, closed access

Congratulations! You’ve just found your first exoplanet transiting a star. You may wonder if there are any additional planets in that stellar system. This information is important in deciding which targets are worth spending more time observing at the telescope. Can we use the known stellar and planet properties to predict whether more transiting planets may be found?

The authors of today’s paper investigated if there were any differences between the properties of systems with only one known transiting exoplanet (single planet systems) and systems where multiple transiting exoplanets orbit the same star (multi-planet systems).

Selecting an unbiased sample

To select their sample of stars, the authors took all the planet candidates from the California Kepler Survey, removed known false positives (signals since proven not to be due to a planet) and non-planetary sized candidates (> 22.4 Earth radii). They also filtered out weak signals (SNR < 10 ) and stars fainter than Kepler magnitude 14.2 to ensure observed differences are astrophysical rather than a selection bias. If only one planet in a multi-planet system survived these cuts then the entire system was removed. The authors also wanted to make sure any additional planets would likely have been detected, so they filtered out systems where a hypothetical planet with 1.5 Earth radii and a 30 day period would not be detected with a SNR of 10.

After these cuts, the authors were left with 376 planets in single-planet systems and 166 multi-planet systems hosting 426 planets. They found no significant difference in the stellar magnitude distribution between single and multi-planet systems, removing the likelihood of bias.

Any difference in the stars?

Figure 1 compares stellar properties between single and multi-planet systems. Each property has a p value which tests whether the difference between the single and multi-planet populations is statistically significant or whether the difference is likely to occur by chance. The middle row of Figure 1 shows a p value of 0.29 for stellar metallicity, meaning there is a 29% probability the difference between the single and multi-planet populations could be due to chance, so the difference is not statistically significant. In fact, no significant difference in stellar mass, metallicity or rotational velocity is found between single and multi-planet systems. Given a random star from the sample, we could not say whether it was more likely to have one or more planets.

Figure 1: Comparison between single (red) and multi-planet (blue) systems for distributions of stellar mass (top row), metallicity (middle row) and rotational velocity (bottom row). Histogram plots are on the lefthand side and cumulative distribution functions on the right. P values indicate no significant difference between single and multi-planet systems. Figure 5 from today’s paper.

Can planets tell us if they are lonesome?

After finding no significant difference in stellar properties between single and multi-planet systems, the authors investigated planetary properties, specifically planet radius and orbital period. The top panel of Figure 2 shows that both single and multi-planet systems display the radius valley, where there is a lack of planets with ~ 1.8 Earth radii. While there are differences in the distribution of planets around the radius gap, there aren’t enough sub-Neptune planets near the valley for the difference to be statistically significant. The radius valley is also seen in multi-planet systems separated into different numbers of transiting planets (middle row of Figure 2).

Figure 2: Comparing planet radii. Top: histogram (left) and cumulative distribution function (right) comparing planet radii between single (red) and multi-planet (blue) systems. Middle: same as above, but comparing multi-planet systems (black) with different numbers of transiting planets, 2 planets in green, 3 in blue and 4 + in purple. Bottom: zoom in of top row. Figure 8 from today’s paper.

Looking at the orbital period distribution for hot-Jupiters (Jupiter-size planets with periods < 10 days) no multi-planet systems are found in this sample, agreeing with previous studies on the rarity of multi-planet systems containing a hot-Jupiter. Focusing on sub-Neptune planets (< 4 Earth radii), more single systems are seen at periods less than 3 days than in multi-planet systems, shown in Figure 3. Beyond 10 days, similar distributions of sub-Neptunes in single and multi-planet systems are seen, with possibly a slight preference for multi-planet systems. Overall, the p value of 0.001 indicates that differences between single and multi-planet system orbital periods are significant (it is 99.9% likely not due to chance). As short period planets are more likely to pass in front of their star, from our view on Earth, could geometry alone explain the higher numbers of single planets? Apparently single transiting planets, especially those transiting at higher inclinations, may in fact be in multi-planet systems but planets further out do not transit. However, if this was the case, then we might expect differences between 2, 3 and 4+ planet systems, which we do not see.

Figure 3: Top panel: Distribution of orbital periods for sub-Neptune single (red) and multi-planet systems (blue). Bottom panel: distribution of orbital periods for all planets in multi-planet systems, same colours as in Figure 2. Figure 9 from today’s paper.


This paper suggests that you cannot differentiate between single and multi-planet systems using stellar properties. If your planet has a short period, being a hot-Jupiter with period < 10 days or a sub-Neptune with period < 3 days, then it is more likely to be a single planet system. If your planet is a sub-Neptune with period > 3 days, then you cannot use this to predict whether it is in a multi-planet transiting system. 

These results question whether the ‘Kepler dichotomy’ (the idea that single and multi-planet systems observed by Kepler are fundamentally different populations with very different system properties) does really exist. However, investigation of whether other planet properties (e.g. eccentricity and mass) vary between single and multi-planet systems is still needed.

About Emma Foxell

I am a PhD student at the University of Warwick. My project involves searching for transiting exoplanets around bright stars using telescopes on the ground. Outside of astronomy, I enjoy rock climbing and hiking.

Leave a Reply