POSEIDON II: The Anti-Aligned Orbit of the Warm Neptune TOI-1710 A b Facilities: WIYN (NEID), TESS, TNG (HARPS-N), Keck:I (HIRES), OHP:1.93m (SOPHIE), APF (Levy), MAST. astropy ( 5 ; 3 ; 4 ) , batman ( 40 ) , celerite ( 26 ) , ceres ( 13 ) , dynesty ( 62 ) , ironman ( 21 ; 23 ) , juliet ( 20 ) , lightkurve ( 44 ) , lofti_gaia ( 51 ) , matplotlib ( 36 ) , numpy ( 69 ) , radvel ( 27 ) , rmfit ( 65 ) , scipy ( 70 ) , serval ( 75 ) , zaspe ( 14 ) . Juan I. Espinoza-Retamal, Hareesh Bhaskar, Joshua N. Winn, Cristobal Petrovich, Rafael Brahm, Caleb Lammers, Guðmundur Stefánsson, Elise Koo, Andrés Jordán, Felipe I. Rojas Abstract We present an observation of the Rossiter-McLaughlin effect for the warm-Neptune system TOI-1710 obtained with the NEID spectrograph on the WIYN 3.5 m telescope. These observations reveal that the planet orbits in the opposite direction to the stellar spin, with a sky-projected obliquity λ = 179 ± 19 ∘ \lambda=179\pm 19^{\circ} . Combined with information about the rotation period of the host star, we measure a true obliquity of ψ = 158 − 13 + 11 ∘ \psi=158_{-13}^{+11}\,{}^{\circ} . The host star has an M-dwarf companion at a separation of ∼ 3600 \sim 3600 au, but this companion is too distant to be solely responsible for misaligning the warm Neptune. The host star also shows a long-term radial velocity trend, indicative of a companion at intermediate separations. We show that such a companion can dynamically couple the warm Neptune to the distant M dwarf, enabling the transfer of inclination from the wide binary orbit to the planetary orbit. Assuming this scenario is correct, we predict the intermediate companion is a ∼ 5 ​ M J \sim 5\,M_{J} planet on a ∼ 15 \sim 15 au orbit that is nearly aligned with the transiting planet’s orbit. I Introduction Stellar obliquities, the angles between stellar spin and planetary orbits, are probes of the dynamical histories of planetary systems ( 1 e.g.,) . Measurements based on the Rossiter-McLaughlin (RM) effect ( 57 ; 47 ) have revealed a wide diversity of system architectures, ranging from well-aligned to strongly misaligned and even retrograde configurations ( 34 ; 71 ; 22 ; 73 ; 12 ; 58 e.g.,) . Misalignments are commonly interpreted as signatures of dynamical processes such as high-eccentricity migration, yet their origin remains debated, particularly for lower-mass planets. Most obliquity measurements have focused on short-period giant planets because they are observationally favorable, but they may not be representative of smaller and more commonly occurring exoplanets ( 35 ; 6 e.g.,) . Recent advances in radial velocity (RV) precision have begun to allow obliquity studies to be extended to Neptune-sized planets ( 65 ; 10 ; 11 ; 32 ; 72 ; 66 ; 55 e.g.,) . Expanding the sample of obliquity measurements in this regime is essential to assess whether the trends observed for giant planets also apply to the more common population of smaller planets. The POSEIDON survey aims to characterize the obliquity distribution of transiting Neptunes and use it to constrain their dynamical origins ( 24 ) . As part of this effort, here we present observations of the RM effect and an obliquity measurement for the warm-Neptune system TOI-1710 ( 39 ) . TOI-1710 is a wide binary composed of a G-type primary star and an M-dwarf companion separated by \sim 3600 au ( 18 ) . A transiting Neptune around the primary was identified by 39 , and its properties were refined by subsequent analyses ( 50 ; 56 ) . The transiting planet has an orbital period of \sim 24 d, a Neptune-like mass of \sim 20\,M_{\oplus} , and a radius of \sim 5\,R_{\oplus} , placing the planet in the “Neptune savanna” ( 10 ) . II Observations II.1 NEID Transit Spectroscopy We observed one transit of TOI-1710 A b using the NEID spectrograph ( 59 ) installed on the 3.5 m WIYN telescope at Kitt Peak Observatory in Arizona. NEID is an echelle spectrograph that covers the wavelength range of 380-930 nm at a resolving power of R\approx 110,\!000 . The transit was observed on UTC October 25, 2025, between 05:39 and 12:26. We obtained 37 spectra spanning the transit with exposure times of 600 s. The spectra were processed with the NEID Data Reduction Pipeline version 1.4.2 ( 8 ) , which resulted in final spectra with a median signal-to-noise ratio of 68 per pixel at 550 nm. We extracted precise RVs using the serval template-matching code ( 75 ) adapted for NEID by 65 . The resulting RVs have a median uncertainty of 1.0 m s -1 , derived from the order-by-order uncertainties. The NEID RVs are shown in Figure 1 and are tabulated in Appendix A . Figure 1 : Observations of TOI-1710 A. (a) NEID RVs, after subtracting a long-term linear trend (purple) along with the best-fit model including the RM effect (red curve) and the associated confidence intervals (1, 2, and 3\sigma , shaded red). Residuals are shown below. (b) Out-of-transit RVs versus orbital phase, along with the best-fit model (the confidence intervals are too small to be seen clearly). Residuals are shown below. (c) RV residuals versus time, after subtracting the signal of TOI-1710 A b. The red line is a linear RV trend, which is detected with 8 \sigma confidence. (d) Transit photometry from TESS (green) along with the best-fit model (red). The darker points are time-averaged data. Residuals are shown below. (e) Deviations between measured transit times and the best-fit constant period model as a function of time. The vertical dashed line indicates the transit that was observed spectroscopically with NEID. II.2 TESS Photometry We retrieved the TESS light curves for Sectors 19, 20, 26, 40, 53, 59, 60, 73, and 79 from the Mikulski Archive for Space Telescopes using the lightkurve package ( 44 ) . We opted for the light curves with 120 s sampling as processed by the Science Processing Operations Center pipeline ( 38 ) , which corrects for pointing and focus-related instrumental signatures, discontinuities resulting from radiation events in the CCD detectors, outliers, and contributions to the recorded flux from nearby stars. The TESS transit observations, along with the best model, are shown in Figure 1 . II.3 Archival Spectroscopy To determine the spectroscopic parameters of the orbit, we used archival RV observations: 30 from SOPHIE ( 39 ) , 85 from HARPS-N ( 50 ) , 178 from the APF, and 16 from HIRES ( 56 ) . We also used archival Keck/HIRES spectra 1 1 1 https://koa.ipac.caltech.edu/cgi-bin/KOA/nph-KOAlogin and processed them with the ceres pipeline ( 13 ) for stellar spectral characterization. III Stellar Parameters We derived the stellar parameters of TOI-1710 A following the iterative procedure described by 15 . First, we determined the atmospheric parameters from the HIRES spectrum using zaspe ( 14 ) . We then estimated the physical parameters by fitting the star’s spectral energy distribution (based on broadband photometry) to synthetic magnitudes from the PARSEC isochrones ( 17 ) and the Gaia DR3 parallax ( 28 ) . The resulting mass and radius were used to calculate \log g , which was then held fixed in a new iteration of the spectroscopic analysis. We repeated this process until convergence in \log g . The final stellar parameters are listed in Appendix B . IV Photometric Analysis In order to refine the orbital ephemeris and to look for possible transit timing variations (TTVs), we analyzed the TESS data with the juliet code ( 20 ) . We modeled the transits with batman ( 40 ) and included a quasi-periodic (QP) Gaussian Process (GP) with celerite ( 26 ) to remove periodic modulation related to the rotational period of TOI-1710 A. We placed broad Gaussian priors on each transit midpoint (width of 1 day) based on the ephemeris of 39 . As shown in Figure 1 , the transit times from Sectors 20, 26, and 40 deviate from the best-fit strictly periodic model by approximately 2.0, 4.0, and 2.5 minutes, respectively. The associated statistical significances are 2.4\sigma , 3.6\sigma , and 2.8\sigma . Although formally significant, we regard the evidence for TTVs as suggestive, not conclusive. The TESS light curves exhibit structured, time-correlated noise that may have caused the timing uncertainties to be underestimated. Additional transit observations will be valuable for determining whether TOI-1710 A b exhibits TTVs related to the presence of companion planets. From the TTV analysis, we obtained an improved orbital ephemeris and a detrended TESS light curve, which we used in the analysis described below. Finally, we note that the posterior for the QP-GP period, 26\pm 16 d, is consistent with previous estimates of the stellar rotation period. 39 reported P_{\rm rot}=22.5\pm 2.0 d based on periodic signals identified in multiple stellar activity indicators (e.g., chromospheric emission indexes) and calibrated activity-rotation relations. Consistently, 50 also found evidence for variability on similar timescales ( \sim 20 – 24 d) in ground-based photometry and modeled the RVs using a QP-GP with a characteristic timescale consistent with stellar rotation. Table 1 : Summary of priors and posteriors of the ironman fit. We have adopted the subindex “A” and “B” for the primary its binary companion, and “b” for the transiting planet. Parameter Description Prior Posterior \psi True stellar obliquity ( ∘ ) \cdots 158_{-13}^{+11} \lambda Sky-projected stellar obliquity ( ∘ ) \mathcal{U}(0,360) 179\pm 19 v Equatorial velocity (km s -1 ) \cdots 2.2\pm 0.2 v\sin{i_{A}} Projected rotational velocity (km s -1 ) \cdots 2.1\pm 0.2 \cos{i_{A}} Cosine of stellar inclination \mathcal{U}(-1,1) 0.0\pm 0.3 i_{A} Stellar inclination ( ∘ ) \cdots 91_{-19}^{+18} P_{\rm rot} Stellar rotational period (d) \mathcal{G}(22.5,2.0) 21.3\pm 1.8 M_{A} Stellar mass ( M_{\odot} ) \mathcal{G}(1.023,0.015) 1.020\pm 0.015 R_{A} Stellar radius ( R_{\odot} ) \mathcal{G}(0.93,0.01) 0.93\pm 0.01 \rho_{A} Stellar mean density (g cm -3 ) \cdots 1.77\pm 0.06 P Orbital period (d) \mathcal{U}(24.28327,24.28346) 24.28336\pm 0.00001 t_{0} Transit midpoint (BJD) \mathcal{U}(2458812.6757,2458812.6833) 2458812.6797\pm 0.0004 b Impact parameter \mathcal{U}(-1,1) 0.00\pm 0.12 i_{b} Orbital inclination ( ∘ ) \cdots 90.01\pm 0.18 R_{b}/R_{A} Radius ratio \mathcal{U}(0,1) 0.0497\pm 0.0003 K RV semiamplitude (m s -1 ) \mathcal{U}(0,1000) 4.2\pm 0.5 \sqrt{e}\sin{\omega} Eccentricity parameter – sine component \mathcal{U}(-1,1) -0.15\pm 0.05 \sqrt{e}\cos{\omega} Eccentricity parameter – cosine component \mathcal{U}(-1,1) 0.09_{-0.20}^{+0.15} e_{b} Eccentricity \cdots 0.05_{-0.02}^{+0.04} \omega_{b} Argument of periastron ( ∘ ) \cdots -62_{-64}^{+36} a_{b}/R_{A} Scaled semimajor axis \cdots 38.0_{-0.5}^{+0.4} R_{b} Planet radius ( R_{\oplus} ) \cdots 5.06\pm 0.06 a_{b} Semimajor axis (au) \cdots 0.1652\pm 0.0008 M_{b} Planet mass ( M_{\oplus} ) \cdots 19.1\pm 2.3 \rho_{b} Planet density (g cm -3 ) \cdots 0.8\pm 0.1 \dot{\gamma} RV slope (m s -1 d -1 ) \mathcal{U}(-1,1) -0.016\pm 0.002 q_{1}^{\rm NEID} NEID linear limb darkening parameter \mathcal{U}(0,1) 0.53\pm 0.29 q_{2}^{\rm NEID} NEID quadratic limb darkening parameter \mathcal{U}(0,1) 0.57_{-0.34}^{+0.29} \beta Intrinsic stellar line width (km s -1 ) \mathcal{G}(4.7,2.0) 3.5_{-2.0}^{+2.4} \gamma_{\rm NEID} NEID RV offset (m s -1 ) \mathcal{U}(-500,500) 24\pm 3 \sigma_{\rm NEID} NEID RV jitter (m s -1 ) \mathcal{LU}(10^{-3},100) 1.1\pm 0.3 \gamma_{\rm SOPHIE} SOPHIE RV offset (m s -1 ) \mathcal{U}(-39000,-38000) -38860.1_{-1.3}^{+1.4} \sigma_{\rm SOPHIE} SOPHIE RV jitter (m s -1 ) \mathcal{LU}(10^{-3},100) 5.9_{-0.8}^{+1.0} \gamma_{\rm HIRES} HIRES RV offset (m s -1 ) \mathcal{U}(-500,500) -3.2\pm 1.3 \sigma_{\rm HIRES} HIRES RV jitter (m s -1 ) \mathcal{LU}(10^{-3},100) 4.6_{-0.9}^{+1.2} \gamma_{\rm HARPS-N} HARPS-N RV offset (m s -1 ) \mathcal{U}(-500,500) -4.9\pm 0.6 \sigma_{\rm HARPS-N} HARPS-N RV jitter (m s -1 ) \mathcal{LU}(10^{-3},100) 4.1_{-0.3}^{+0.4} \gamma_{\rm APF} APF RV offset (m s -1 ) \mathcal{U}(-500,500) 1.1\pm 0.6 \sigma_{\rm APF} APF RV jitter (m s -1 ) \mathcal{LU}(10^{-3},100) 7.9_{-0.4}^{+0.5} q_{1}^{\rm TESS} TESS linear limb darkening parameter \mathcal{U}(0,1) 0.20_{-0.06}^{+0.08} q_{2}^{\rm TESS} TESS quadratic limb darkening parameter \mathcal{U}(0,1) 0.60_{-0.17}^{+0.2} \sigma_{\rm TESS} TESS photometric jitter (ppm) \mathcal{LU}(1,5\times 10^{7}) 94_{-77}^{+33} Note. — \mathcal{U}(a,b) denotes a uniform prior with a start value a and end value b . \mathcal{G}(\mu,\sigma) denotes a normal prior with mean \mu , and standard deviation \sigma . \mathcal{LU}(a,b) denotes a log-uniform prior with a start value a and end value b . V Obliquity Determination To constrain the stellar obliquity, we performed a joint analysis of the observations described in Section II using the ironman 2 2 2 https://github.com/jiespinozar/ironman package ( 21 ; 23 ) . ironman is a versatile python package that can jointly fit the RM effect, Keplerian RVs, and transit photometry, by joining other packages: rmfit ( 65 ) for the RM effect modeling, batman ( 40 ) for the transit light curves, radvel ( 27 ) for the Keplerian RVs, and dynesty ( 62 ) to sample the posteriors. We adopt the stellar rotation period reported by 39 to derive the true obliquity using the parametrization from 65 available in ironman . This parametrization performs the correct accounting for the correlation between v\sin{i} and the equatorial velocity of the star ( 46 see) . In this analysis we worked with the TESS light curve that was detrended in Section IV , and considered only within 10 hours of the transits to reduce computational cost. We included independent jitter terms for each photometric and RV instrument to account for possible systematics. We placed uniform priors on almost all parameters, except for the stellar parameters (mass, radius, rotation period). All priors and results obtained from the posterior distributions are shown in Table 1 . We find that TOI-1710 A b has a nearly circular ( e=0.05_{-0.02}^{+0.04} ) and retrograde orbit with \lambda=179\pm 19\,^{\circ} and \psi=158_{-13}^{+11}\,{}^{\circ} . We also noticed the presence of a long-term RV trend (see Figure 1 ). We modeled it in our joint fit using a linear function of the form \dot{\gamma}\left(t-t_{a}\right) where we arbitrarily chose t_{a}=2459539.2192 , near the middle of the observing baseline. We compared the models with and without the linear trend using the Bayesian evidence ( \log{Z} ). The model with the linear trend is favored by \Delta\log{Z}=18 . The RV slope is \dot{\gamma}=-0.016\pm 0.002 m s -1 d -1 , an 8\sigma detection. Overall, the parameters of the system agree to within 2\sigma with those reported by 39 , 50 , and 56 . VI Discussion VI.1 Search for Companions The catalog of 18 includes a spatially resolved M-dwarf companion to TOI-1710 A, identified based on similar proper motion and parallax. The companion has Gaia DR3 identifier 1116612783096856960, and will be hereafter called TOI-1710 B. The current projected separation is \approx 45^{\prime\prime} , which translates into \approx 3610 au. Although the Gaia parallaxes of the two stars are consistent, their proper motions differ by a small but significant amount, and this relative sky-plane velocity can be used to constrain the binary orbit. For this purpose, we used the lofti_gaia package ( 51 ) , which constrains binary orbits using Gaia astrometry by fitting the relative position and proper motion of the system at a single epoch. In brief, the code generates trial Keplerian orbits, scales and rotates them to match the observed projected separation, and then accepts or rejects them based on how well they reproduce the observed relative velocity. The code requires as inputs the Gaia IDs of both stars and knowledge of their masses. For TOI-1710 A, we used the value reported in Appendix B of 1.023\pm 0.015\,M_{\odot} , while for TOI-1710 B, we used 0.40\pm 0.02\,M_{\odot} , the value reported in the TESS Input Catalog ( 64 ; 63 ) . After 100,000 accepted orbits we obtained a semimajor axis a_{B}=3354^{+1056}_{-1527} au, inclination i_{B}=103_{-8}^{+7}\,{}^{\circ} , and eccentricity e_{B}=0.82^{+0.16}_{-0.15} . The reported parameters for the orbit of TOI-1710 B are the median and 68% credible intervals of the accepted orbits. As the planet’s orbit has i_{b}\approx 90^{\circ} , the minimum angle between the planetary and binary orbital planes is \approx 13\pm 8^{\circ} . We also evaluate the possibility of additional planetary companions. As discussed in Section IV , there is suggestive but inconclusive evidence for TTVs. A search for additional transiting planets using a box least squares algorithm did not reveal any significant signals. In contrast, the RVs show evidence for a long-term trend. This trend cannot be attributed to the wide binary companion, as its expected contribution based on the derived orbital parameters is several orders of magnitude too small. If instead the RV trend is produced by a planetary companion, the planet’s orbital period must be longer than the current RV baseline ( \gtrsim 2.4 yr). The system is therefore a promising target for future astrometric constraints with Gaia ( 52 ; 25 ; 43 e.g.,) , given its proximity and brightness ( d\approx 81 pc, G\approx 9.4 mag). Unfortunately, it was not observed by Hipparcos, and is therefore not included in the Hipparcos-Gaia catalog of accelerating sources ( 16 ) , limiting current astrometric constraints on additional companions. No additional periodic signals are detected in the RVs after removing the signal of TOI-1710 A b and the long-term trend. Figure 2 : Secular evolution of the obliquities of TOI-1710 from four-body secular integrations. The semimajor axis of the intermediate companion varies across panels. The obliquity evolution of TOI-1710 A b (red) and the hypothetical intermediate companion X (blue) are shown. The companion mass is fixed at 9\,M_{J} and the initial mutual inclination with the M-dwarf is i_{XB}=55^{\circ} . The green region marks the 2\sigma constraints on the obliquity of the observed planet, with the fraction of time spent within this region indicated by f_{\rm obs} . Panels (a) and (c) remain within the allowed window for much less time than panel (b). In all cases, the eccentricity of planet b remains nearly the same as the initial value. VI.2 Possible Origins of the Retrograde Orbit How can we explain the retrograde and nearly circular orbit of TOI-1710 A b? Disk torquing One possibility is that the protoplanetary disk was strongly tilted with respect to the stellar spin axis. Such disk misalignments might be caused by gravitational torques from stellar companions ( 7 ; 42 ; 74 e.g.,) . However, the nodal precession timescale of a disk around star A induced by the torque from star B is likely too long compared to the disk dispersal timescale to efficiently excite large stellar obliquities. Using the binary properties described in Section VI.1 and following 42 , we estimate that a protoplanetary disk extending out to 100 au would complete half a precession cycle in \sim 40 Myr, significantly exceeding typical protoplanetary disk lifetimes ( 19 e.g.,) . Alternatively, the stellar obliquity may have been excited through magnetic disk-star interactions during the early accretion phase, in which torques exerted by a magnetically coupled disk can tilt the stellar spin axis ( 41 ; 61 e.g.,) . High-eccentricity tidal migration Secular perturbations from the distant stellar companion or an unseen massive planetary companion might have driven high-eccentricity migration of the Neptune-mass planet, possibly explaining the retrograde orbit of TOI-1710 A b. However, this scenario requires efficient tidal dissipation to circularize the orbit. This seems unlikely, because the expected circularization timescale (given the planet’s relatively wide orbital separation) greatly exceeds the age of the system. 3 3 3 Following 29 and 37 , and adopting a modified tidal quality factor ( 49 e.g.,) of Q^{\prime}=10^{4} , we estimate a tidal circularization timescale of \sim 30 Gyr. Secular inclination cascade driven by the distant star For the rest of this section, we consider a scenario in which the orbit of TOI-1710 A b is tilted through secular interactions with the M-dwarf companion and the additional substellar companion that produces the long-term radial velocity trend. Such a companion, hereafter called planet X, could facilitate the transfer of inclination from the binary orbit to the planetary orbit in a cascading manner. This mechanism is similar to that proposed by 9 for the retrograde system K2-290 ( 33 ; 34 ) . VI.2.1 Numerical Experiments of the Secular Inclination Cascade Driven by the Distant Companion We adopt the currently observed orbital configuration of TOI-1710 A b as the initial condition, namely a semimajor axis of a=0.16 au and a nearly circular orbit ( e=0.05^{+0.04}_{-0.02} ). We also adopt the present-day stellar rotation period of 21.3\pm 1.8 days, because obliquity evolution occurs on timescales of hundreds of Myr after formation, over which the stellar spin is not expected to vary significantly. The known M-dwarf companion, with a semimajor axis of 3354^{+1056}_{-1527} au, is too distant to overcome the planet’s strong coupling to the stellar spin, and therefore cannot by itself generate the extreme obliquity of TOI-1710 A b ( \psi=158^{+11}_{-13}\,\mathrm{deg} ). 4 4 4 The Laplace radius associated with the M-dwarf companion is 0.6 au, placing TOI-1710 A b well inside the regime where spin–orbit coupling is dominant. We therefore construct a hierarchical four-body model comprising the host star, planet b, planet X, and the distant M-dwarf companion, and perform an ensemble of secular simulations to constrain the mass and semimajor axis of planet X (see Appendix C for details). Figure 2 shows three calculations of the secular evolution of the obliquities of planets b and X (i.e., their orbital inclinations relative to the star’s equatorial plane) for different choices of planet X’s semimajor axis ( a_{X} ). In all cases, even though the obliquities are initially very small, planet b’s obliquity evolves to reach the 2\sigma lower limit of \psi=140^{\circ} , consistent with observations, while its eccentricity remains close to the initial value. To quantify the agreement with observations, we define f_{\rm obs} as the fraction of the integration time during which the simulated system satisfies the obliquity constraints. Panels (a) and (b) show simulations in which planets b and X are tightly coupled, allowing them to maintain nearly aligned orbital planes throughout the evolution. In panel (a), the companion lies too close to the host star, leading to delayed obliquity excitation and only brief excursions into the observationally allowed range within the system’s age (low f_{\rm obs} ). In contrast, panel (b) shows a configuration in which the separation of planet X is close to optimal for obliquity excitation; in this case, the system spends nearly half of the integration time within the observed obliquity range. In neither case can the evolution be explained solely by nodal precession of the planetary orbits, for which the maximum \psi would be 2\,i_{\rm XB}\sim 110^{\circ} . Instead, a secular spin–orbit resonance occurs at a_{X}\sim 10 au (Appendix C ), where the stellar spin precession rate becomes comparable to the nodal precession rate induced by the M-dwarf companion. This resonance drives the evolution of the stellar spin vector, enabling the obliquity of planet b to reach the observationally allowed range. Panel (c) illustrates a case in which a secular orbit–orbit resonance arises between the nodal precession of planet X induced by the M-dwarf and that of planet b induced by planet X, leading to chaotic evolution. As in panel (a), the system satisfies the observational constraints for only a small fraction of the integration time ( f_{\rm obs}<0.1 ). VI.3 Predictions for Planet X Having shown that secular inclination excitation can account for both the low eccentricity and high obliquity of the observed planet b, we now combine the radial velocity trend with population-synthesis predictions to further constrain the properties of the hypothesized planet X. The outcome of secular evolution depends sensitively on the mutual inclination between the planetary system and the stellar companion B. We perform an ensemble of secular simulations over a range of choices for the mutual inclination, as well as planet X’s mass and semimajor axis. The most successful simulations occur when the mutual inclination satisfies either 50^{\circ}\lesssim i_{XB}\lesssim 70^{\circ} or 110^{\circ}\lesssim i_{XB}\lesssim 130^{\circ} . For more aligned or anti-aligned configurations ( i_{XB}<50^{\circ} or > 130 ∘ ), the obliquity of planet b is not significantly excited, and for near-polar configurations ( 70^{\circ}20 AU). Such a configuration induces eccentricity oscillations in the intermediate companion, which subsequently drives it close to the observed planet, ultimately destabilizing the system. In our analysis, we ignore simulations which have been marked as unstable. Figure 4 : Results of the ensemble of secular simulations. The semimajor axis of the intermediate companion X is shown on the x-axis, and its mass is shown on the y-axis. The color scale indicates the maximum obliquity attained by the observed planet b. In all simulations, the initial inclination between the planets and the M-dwarf companion is set to 50^{\circ} . Analytical expressions delineating the different dynamical regimes are shown as solid and dashed lines. See the text for a detailed explanation. We use the following initial conditions to run these simulations: a_{b}=0.165 AU, e_{b}=0.05 , e_{X}=0.1 , i_{bX}=2.6^{\circ} , and i_{XB}=56^{\circ} . The initial spin periods of the host star and the observed planet are taken to be 22 days, and 0.5 days respectively. The initial obliquities of the planets are set close to zero ( =0.025 ). The viscous dissipation timescales of the star and the planet are set to 50 years and 0.1 years, respectively. Meanwhile, the love numbers of the star and the observed planet are set to k_{2,0} = 0.028 and k_{2,p}= 0.5, respectively. Figure 4 shows results from our ensemble of simulations. When planet X orbits close to the host star ( a_{X}<7\,\mathrm{au} ), both the eccentricity and obliquity of planet b remain low ( e_{b,\max}\sim 0.05 , \psi_{b,\max}<10^{\circ} ). This region is bound on the right side by the dashed green line , which marks the semimajor axis at which the precession timescale of the intermediate companion due to the M-dwarf companion is comparable to the age of the system ( \sim 10^{9} years). This timescale is given by the ZKL timescale: t_{\mathrm{\rm ZKL,XB}}\equiv\left(\frac{M_{\mathrm{A}}}{M_{\mathrm{B}}}\right)\left(\frac{a_{\mathrm{B}}}{a_{\mathrm{X}}}\right)^{3}\left(1-e_{\mathrm{B}}^{2}\right)^{3/2}P_{X}, (C1) ( 2 e.g.,) . To the left of this line, planet X is effectively decoupled from the M-dwarf. In this regime, the star’s obliquity relative to planet b remains low and incompatible with observations. A second low-excitation regime exists where the intermediate companion is both low-mass ( M_{X}\lesssim 10^{-2} ) and distant ( a_{X}\gtrsim 20\,\mathrm{AU} ). This region is bounded on the left side by the solid green line , where the semimajor axis at which the nodal precession rate of planet b induced by the stellar J_{2} becomes comparable to the nodal precession rate induced by planet X. This condition is obtained by using the expression for Laplace radius of planet b: a_{b}=r_{L}=\left(J_{2}R_{A}^{2}a_{\mathrm{X}}^{3}\frac{M_{A}}{M_{\mathrm{X}}}\right)^{1/5}. (C2) Here, R_{A} is the radius of the host star ( 68 e.g., ) and J_{2} is the star’s gravitational moment: J_{2}=\frac{k_{2,A}}{3}\left(\frac{\Omega_{s,A}}{\Omega_{s,brk}}\right)^{2}, (C3) where \Omega_{s,A} is the spin rate of the host star and \Omega_{s,brk}=\sqrt{GM_{A}/R_{A}^{3}} is the breakup rate. To the right of the solid green line, the stellar J_{2} dominates and the influence of planet X is negligible. This region is inconsistent with observations, as the obliquity remains unexcited. The elevated obliquities near a_{X}\sim 10\,\mathrm{AU} arise from a spin–orbit resonance. This occurs when the stellar spin precession rate (driven by torques from the inner planet) becomes comparable to the nodal precession rate induced by the M-dwarf companion on planet X. In this regime, the intermediate companion and planet b remain strongly coupled and precess together. The stellar spin precession rate induced by planet b is given by \alpha=\frac{k_{A}}{k_{I,A}}\frac{M_{b}}{M_{A}}\left(\frac{R_{A}}{a_{b}}\right)^{3}\Omega_{s,A}, where k_{A}=k_{2,A}/2 is the stellar apsidal motion constant and k_{I,A}=0.08 is the moment-of-inertia constant. For this system, we obtain \alpha=1.8\times 10^{-8}\,\mathrm{yr}^{-1} . A spin–orbit resonance occurs when this rate becomes comparable to the nodal precession rate (i.e., \alpha t_{\rm ZKL,XB}\sim 1 ). In this regime, the stellar spin undergoes large secular variations, allowing high obliquities to be reached. Another distinct region where the obliquity of the observed planet is strongly excited ( \psi_{b,\max}>50^{\circ} ) involves a different secular orbital resonance: the nodal precession rate of planet X due to the M-dwarf matches the precession rate of planet b due to planet X. The contours of the precession-rate ratio \mathcal{R}=1 and \mathcal{R}=10 are shown as solid and dashed red lines, respectively. The analytical estimate for \mathcal{R} is given by ( 31 ) : \displaystyle\mathcal{R} \displaystyle\equiv \displaystyle\frac{t_{\mathrm{ZKL},bX}}{t_{\mathrm{ZKL},XB}} (C4) \displaystyle\simeq \displaystyle\frac{M_{B}}{M_{X}}a_{b}^{-\frac{3}{2}}a_{X}^{\frac{9}{2}}a_{B}^{-3}\frac{1}{(1-e_{B}^{2})^{\frac{3}{2}}}. This condition marks the onset of chaotic behavior, which can drive large obliquity excitation as recently shown by Liu et al. (in prep.). Overall, we can see that there is broad region of parameter space in which the obliquity of the observed planet can be excited to large values, consistent with the observations.