Sub-Jovian Desert in Exoplanet Studies

• Sub-Jovian Desert is the region of close-in exoplanets where Neptune-like planets are scarce, lying between super-Earths and hot Jupiters.
• Research shows its boundaries form triangular, wedge-like shapes in period–mass and period–radius spaces, indicating distinct processes like photoevaporation and tidal disruption.
• Studies integrate host-star properties and irradiation proxies to refine desert limits, highlighting atmospheric escape and migration as key factors in planetary evolution.

The Sub-Jovian Desert—also called the hot Neptune desert, Neptunian desert, or, in some usages, the sub-Jovian/Neptune desert—is the underpopulated region of close-in exoplanet parameter space occupied by planets intermediate between short-period super-Earths/sub-Neptunes and hot Jupiters. It is seen in both the orbital period–planet mass and orbital period–planet radius planes, where very short-period small planets and hot Jupiters are common but Neptune-like to sub-Jovian planets are comparatively rare. Large-sample analyses treat this deficit as astrophysical rather than purely instrumental, and subsequent work has interpreted it as a demographic imprint of atmospheric escape, migration, tidal evolution, and, in some models, the destruction of gas giants (Mazeh et al., 2016, Owen et al., 2018).

1. Definition and observational manifestation

In its classical observational form, the desert is the paucity of planets with Neptune-like masses and radii on very short orbits. Early surveys described it in the period–mass plane as a lack of planets with masses of about $0.03$–$0.3\,M_{\rm Jup}$ below roughly 5–10 days, and in the period–radius plane as a comparable gap for planets with radii of roughly $3$–$10\,R_\oplus$ at short periods (Mazeh et al., 2016). Other descriptions emphasize the most extreme short-period regime, noting a dearth of planets with $P_{\rm orb}\lesssim 2$–4 days and intermediate masses or radii between super-Earths and hot Jupiters (Eigmüller et al., 2019, Smith et al., 2021).

The region is not completely empty. Several studies explicitly stress that the desert is sparsely populated rather than vacant, and one paper recasts it as a “sub-Jovian savanna” containing a small number of conspicuous “giraffe” planets such as LTT 9779 b, TOI-674 b, and WASP-156 b (Kálmán et al., 2023). Work on TOI-2196 b similarly argues that the hottest part of the desert is occupied by a very small number of well-characterized planets rather than being uniformly devoid of objects (Persson et al., 2022).

The significance of the desert lies in its location between two abundant populations. Above it lie hot Jupiters and inflated gas giants; below it lie short-period super-Earths and compact sub-Neptunes. This separation has often been compared to the brown-dwarf desert: the implication is not merely rarity, but that different formation and survival channels may dominate on the two sides of the gap (Mazeh et al., 2016, Owen et al., 2018).

2. Geometric structure and boundary laws

A central empirical result is that the desert is not rectangular. In both the mass and radius projections it has a triangular or wedge-like shape, with an upper boundary and a lower boundary that slope in opposite directions and meet toward very short periods (Mazeh et al., 2016, Owen et al., 2018). In the notation

$$\mathcal{M}=\log_{10}(M_{\rm p}/M_{\rm Jup}),\quad \mathcal{R}=\log_{10}(R_{\rm p}/R_\oplus),\quad \mathcal{P}=\log_{10}(P_{\rm orb}/{\rm d}),$$

the classical period-based boundary fits are:

Plane	Upper boundary	Lower boundary
Period–mass	$\mathcal{M}=-0.99\,\mathcal{P}+0.18$	$\mathcal{M}=0.98\,\mathcal{P}-1.85$
Period–radius	$$\mathcal{R}=-(0.31\pm0.12)\mathcal{P}+(1.19\pm0.06)$$	$\mathcal{R}=(0.67\pm0.06)\mathcal{P}-(0.01\pm0.04)$

These fits imply approximate scalings of $0.3\,M_{\rm Jup}$0 and $0.3\,M_{\rm Jup}$1 for the upper edge, with $0.3\,M_{\rm Jup}$2 and $0.3\,M_{\rm Jup}$3 for the lower edge (Mazeh et al., 2016). Combining the two upper boundaries yields the shallow mass–radius relation

$0.3\,M_{\rm Jup}$4

for $0.3\,M_{\rm Jup}$5 (Mazeh et al., 2016).

Later work argued that the period-only description is incomplete because it suppresses host-star dependence. One approach replaces $0.3\,M_{\rm Jup}$6 by the mass ratio $0.3\,M_{\rm Jup}$7 and replaces orbital period by equilibrium temperature $0.3\,M_{\rm Jup}$8, on the grounds that migration depends on stellar mass while inflation depends more directly on irradiation than on period (Eigmüller et al., 2019). Another reformulation uses incident stellar flux $0.3\,M_{\rm Jup}$9 instead of period and finds power-law upper and lower boundaries in the $3$0 and $3$1 planes; in flux–radius space the preferred fits are

$3$2

with analogous power laws in flux–mass space (Magliano et al., 2024). This suggests that the desert is better understood as an irradiation-dependent depletion region than as a universal gap at fixed period.

3. Physical interpretations

A widely used two-process interpretation assigns different mechanisms to the two edges. In this picture, the lower boundary is sculpted by photoevaporation of low-mass H/He-rich planets, while the upper boundary is set by tidal disruption during high-eccentricity migration of gas giants (Owen et al., 2018). Photoevaporation is treated as an early-time process, with stellar high-energy emission strong for about $3$3 Myr, and the lower edge is reproduced if the low-mass planet population has a maximum core mass of about $3$4–$3$5 (Owen et al., 2018). For the upper boundary, Roche-limit arrival and tidal circularization yield a natural minimum surviving period for gas giants; one high-eccentricity migration model finds that the observed upper boundary is consistent with a stellar tidal quality factor of order $3$6 (Matsakos et al., 2016).

Several demographic studies favor an irradiation-driven origin for at least part of the structure. One analysis finds that the desert boundary depends on stellar parameters in the order $3$7, $3$8, then $3$9, while showing no significant dependence on current tidal forcing, planetary surface gravity, or Roche-lobe filling factor; the same work concludes that the distributions are most compatible with the dominant role of photoevaporation (Szabó et al., 2019). A later study using a “golden sample” of 650 exoplanets reports that the upper boundary itself is multiform: in the $10\,R_\oplus$0 plane it is marked by inflated hot Jupiters, whereas in the $10\,R_\oplus$1 plane it is traced by normal hot Jupiters; the same paper also reports chemical substructures, with the boundary more sensitive to volatiles and alpha-elements than to refractory elements (Szabó et al., 2023).

More recent work has proposed “top-down” channels in which at least some desert occupants are remnants of hot Jupiters. One Roche-lobe-overflow model argues that lossy RLO, with $10\,R_\oplus$2, can strip away essentially the entire envelope and backfill the entire width of the desert with dense remnants; it further predicts that LTT 9779 b should be tidally decaying at roughly $10\,R_\oplus$3 ms/yr if it is such a remnant (Hallatt et al., 26 Sep 2025). A related dynamical study finds that RLO clears out companions inside orbital periods $10\,R_\oplus$4 days, whereas companions beyond that range can survive, providing a way to distinguish RLO from destruction during high-eccentricity migration (Liveoak et al., 21 Apr 2026). An alternative dynamical-tide model attributes the hot Jupiter pile-up, Neptune ridge, and Neptune desert to the outcome of tidally excited $10\,R_\oplus$5-modes: shallow, diffusive shocks circularize and preserve envelopes, whereas deeper shocks drive outflows and leave stripped cores in the desert (Zanazzi et al., 18 Jun 2026).

Taken together, these studies favor a multi-channel interpretation. The lower edge is commonly linked to envelope loss, the upper edge to migration and tidal survival, while some individual “desert dwellers” may be either survivors of extreme irradiation or remnants of more massive progenitors.

4. Benchmark planets and boundary tracers

The desert is mapped not only statistically but also through a small number of unusually well-characterized planets.

Planet	Representative properties	Role in desert studies
TOI-2196 b	$10\,R_\oplus$6 d; $10\,R_\oplus$7; $10\,R_\oplus$8; $10\,R_\oplus$9 K	Rare hot desert occupant
NGTS-14Ab	$P_{\rm orb}\lesssim 2$0 d; $P_{\rm orb}\lesssim 2$1; $P_{\rm orb}\lesssim 2$2; $P_{\rm orb}\lesssim 2$3 K	Survivor inside the desert
NGTS-5b	$P_{\rm orb}\lesssim 2$4 d; $P_{\rm orb}\lesssim 2$5; $P_{\rm orb}\lesssim 2$6	Upper-boundary object; host-star dependent classification
K2-60b	$P_{\rm orb}\lesssim 2$7 d; $P_{\rm orb}\lesssim 2$8; $P_{\rm orb}\lesssim 2$9; $$\mathcal{M}=\log_{10}(M_{\rm p}/M_{\rm Jup}),\quad \mathcal{R}=\log_{10}(R_{\rm p}/R_\oplus),\quad \mathcal{P}=\log_{10}(P_{\rm orb}/{\rm d}),$$0 g cm$$\mathcal{M}=\log_{10}(M_{\rm p}/M_{\rm Jup}),\quad \mathcal{R}=\log_{10}(R_{\rm p}/R_\oplus),\quad \mathcal{P}=\log_{10}(P_{\rm orb}/{\rm d}),$$1	Compact boundary tracer
TOI-3568 b	$$\mathcal{M}=\log_{10}(M_{\rm p}/M_{\rm Jup}),\quad \mathcal{R}=\log_{10}(R_{\rm p}/R_\oplus),\quad \mathcal{P}=\log_{10}(P_{\rm orb}/{\rm d}),$$2 d; $$\mathcal{M}=\log_{10}(M_{\rm p}/M_{\rm Jup}),\quad \mathcal{R}=\log_{10}(R_{\rm p}/R_\oplus),\quad \mathcal{P}=\log_{10}(P_{\rm orb}/{\rm d}),$$3; $$\mathcal{M}=\log_{10}(M_{\rm p}/M_{\rm Jup}),\quad \mathcal{R}=\log_{10}(R_{\rm p}/R_\oplus),\quad \mathcal{P}=\log_{10}(P_{\rm orb}/{\rm d}),$$4; $$\mathcal{M}=\log_{10}(M_{\rm p}/M_{\rm Jup}),\quad \mathcal{R}=\log_{10}(R_{\rm p}/R_\oplus),\quad \mathcal{P}=\log_{10}(P_{\rm orb}/{\rm d}),$$5 K	“Bridge” planet in the transition zone

TOI-2196 b is especially important because it occupies the ultra-irradiated regime with both a precise mass and radius. Its density of about $$\mathcal{M}=\log_{10}(M_{\rm p}/M_{\rm Jup}),\quad \mathcal{R}=\log_{10}(R_{\rm p}/R_\oplus),\quad \mathcal{P}=\log_{10}(P_{\rm orb}/{\rm d}),$$6 is roughly twice Neptune’s, yet interior modeling still allows a thin primordial H/He envelope with an H/He mass fraction of roughly 0.4% to 3%, with a mean value of 0.7% (Persson et al., 2022). The same work argues that for planets with $$\mathcal{M}=\log_{10}(M_{\rm p}/M_{\rm Jup}),\quad \mathcal{R}=\log_{10}(R_{\rm p}/R_\oplus),\quad \mathcal{P}=\log_{10}(P_{\rm orb}/{\rm d}),$$7 K the hot Neptune desert may split into a hot sub-Neptune desert at roughly 1.8–3 $$\mathcal{M}=\log_{10}(M_{\rm p}/M_{\rm Jup}),\quad \mathcal{R}=\log_{10}(R_{\rm p}/R_\oplus),\quad \mathcal{P}=\log_{10}(P_{\rm orb}/{\rm d}),$$8 and a sub-Jovian desert at roughly 5–12 $$\mathcal{M}=\log_{10}(M_{\rm p}/M_{\rm Jup}),\quad \mathcal{R}=\log_{10}(R_{\rm p}/R_\oplus),\quad \mathcal{P}=\log_{10}(P_{\rm orb}/{\rm d}),$$9, although the sample is explicitly said to be too small to establish this as a secure physical bimodality (Persson et al., 2022).

NGTS-14Ab provides a contrasting example of a planet inside the desert that appears to have retained part of its primordial atmosphere rather than being a bare remnant. Its measured radius and density are described as broadly compatible with a significant gaseous envelope, and comparison to theoretical boundaries places it near the lower desert boundary expected from photoevaporation models (Smith et al., 2021). TOI-3568 b occupies the transition between two populations in both mass–period and radius–period space; despite receiving one of the highest EUV luminosities among planets with $\mathcal{M}=-0.99\,\mathcal{P}+0.18$0, it lies above the $\mathcal{M}=-0.99\,\mathcal{P}+0.18$1 Gyr evaporation-lifetime contour, making it a useful test case for survival against complete stripping (Martioli et al., 2024).

Boundary objects can look very different depending on whether mass or radius is used. NGTS-5b lies within the mass/period desert as defined in that study, but in the radius/period plane it is well above the boundary and sits in the inflated hot-Jupiter population (Eigmüller et al., 2019). K2-60b probes the desert from the compact side: it is denser than expected for its irradiation level and has a radius anomaly of $\mathcal{M}=-0.99\,\mathcal{P}+0.18$2, making it a counterpoint to the inflated giants that often dominate very short-period transit samples (Eigmüller et al., 2016).

5. Stellar dependence and alternative coordinates

A persistent result of desert studies is that the apparent boundary depends on host-star properties. In one period-based analysis, planets around cooler stars can exist at substantially shorter periods than analogous planets around hotter stars; in the main test region, more than 60% of planets around $\mathcal{M}=-0.99\,\mathcal{P}+0.18$3 stars lie at $\mathcal{M}=-0.99\,\mathcal{P}+0.18$4 days, compared with only about 10% around hotter stars (Szabó et al., 2019). Another study identifies the strongest boundary sensitivity with effective temperature and stellar radius, finds stellar mass dependence weaker, and treats the overall distribution as composed of distinct large-planet and small-planet clusters with different regressions against stellar properties (Szabó et al., 2023).

This dependence motivates replacing period with a direct irradiation proxy. One formulation writes

$\mathcal{M}=-0.99\,\mathcal{P}+0.18$5

making explicit that equal orbital periods correspond to different stellar fluxes for different hosts (Magliano et al., 2024). In that flux-based analysis, the desert appears most clearly above about $\mathcal{M}=-0.99\,\mathcal{P}+0.18$6 in the radius plane and above about $\mathcal{M}=-0.99\,\mathcal{P}+0.18$7 in the mass plane (Magliano et al., 2024).

The quantitative reclassification is substantial. Of 221 planets that would be labeled “hot Neptunes” in the traditional $\mathcal{M}=-0.99\,\mathcal{P}+0.18$8–$\mathcal{M}=-0.99\,\mathcal{P}+0.18$9 diagram, 194 lie outside the irradiation desert in flux space; only 27 remain inside (Magliano et al., 2024). This suggests that period-only definitions can overstate the apparent scarcity of some planets by conflating objects with very different irradiation environments.

Late-type hosts are particularly important in this respect. NGTS-5b orbits a K2V star, and its interpretation shifts when the host is treated correctly: ignoring stellar type makes it look deeper in the desert, whereas using $\mathcal{M}=0.98\,\mathcal{P}-1.85$0 and $\mathcal{M}=0.98\,\mathcal{P}-1.85$1 versus $\mathcal{M}=0.98\,\mathcal{P}-1.85$2 moves it toward the upper boundary (Eigmüller et al., 2019). The broader implication is that empirical desert boundaries are not strictly universal in simple $\mathcal{M}=0.98\,\mathcal{P}-1.85$3 or $\mathcal{M}=0.98\,\mathcal{P}-1.85$4 space.

6. Observation, atmospheric escape, and current limitations

Because the desert is sparse, population inference is sensitive to confirmation quality and atmospheric diagnostics. Multi-band transit vetting of short-period Kepler candidates found that KOI 439.01 and KOI 732.01 are likely planets within the desert, while KOI 531.01 is likely a false positive, demonstrating that chromatic false positives can materially affect how densely the region appears populated (Colón et al., 2015). This is especially relevant because desert occupancy is often inferred from small-number statistics.

Atmospheric escape measurements are increasingly central. One JWST study uses metastable helium at 1.083 $\mathcal{M}=0.98\,\mathcal{P}-1.85$5m as a tracer of weak outflows and finds that a single transit of GJ 436 b with NIRSpec/G140H is sensitive to mass-loss rates two orders of magnitude lower than ground-based limits, down to about $\mathcal{M}=0.98\,\mathcal{P}-1.85$6 rather than only $\mathcal{M}=0.98\,\mathcal{P}-1.85$7 (Santos et al., 2023). The same work shows that unresolved helium absorption introduces a strong degeneracy between $\mathcal{M}=0.98\,\mathcal{P}-1.85$8 and $\mathcal{M}=0.98\,\mathcal{P}-1.85$9 in a 1D isothermal Parker-wind interpretation, so more sophisticated hydrodynamic models are needed to turn helium detections into robust escape rates (Santos et al., 2023).

Future photometric programs are framed explicitly around the desert. Simulations combining TESS, PLATO/NCAM, and the three Ariel filters VISPhot, FGS1, and FGS2 indicate that Ariel’s three-band combination can reach an internal precision of $$\mathcal{R}=-(0.31\pm0.12)\mathcal{P}+(1.19\pm0.06)$$0 on planetary radius for the selected “giraffe” planets, while the multi-mission time baseline would extend over decades and could reveal time evolution in apparent planetary radius or transit morphology (Kálmán et al., 2023). These programs are intended to discriminate between ongoing escape, long-term survival, and structurally protected atmospheres.

The main limitation remains sample size. In the most relevant $$\mathcal{R}=-(0.31\pm0.12)\mathcal{P}+(1.19\pm0.06)$$1 K regime, the TOI-2196 study identifies only a few planets with well-measured properties—TOI-2196 b, K2-100 b, TOI-849 b, LTT 9779 b, together with the radius-only objects K2-278 b and Kepler-644 b (Persson et al., 2022). A plausible implication is that current arguments about bimodality, remnant-core formation, or flux-defined reclassification remain provisional until the sample of precise masses, radii, and atmospheric constraints becomes substantially larger. At present, the Sub-Jovian Desert is best understood not as a single empty gap, but as a structured and still actively reinterpreted boundary region in close-in planet demographics.