Climatic distance to cultivar origin predicts Deglet Noor date quality across two continents

Chouaib Selkh , Samia Bissati, Abdelmadjid Chehma

Background. Premium-grade cultivars of terroir-controlled crops
are tied to specific climatic origins, yet the quantitative
relationships linking climate, soil and quality remain poorly
resolved across whole production landscapes. Deglet Noor date
palm — the historically dominant premium cultivar of North
African oases, named for the optical translucency of its
top-grade fruits — exemplifies this knowledge gap and is
increasingly threatened by climate change.

Methods. Using 35 years of ERA5 reanalysis (1990–2024) on a
controlled-cultivar panel of thirteen Deglet Noor sites spanning
the Algerian Septentrional Sahara and the Tunisian
Jerid–Nefzaoua, we tested two hypotheses: (i) at the event
scale, that orographic basin topography amplifies thermal
extremes and depresses diurnal temperature range during fruit
maturation; and (ii) at the site scale, that quality declines
non-linearly with a four-dimensional climatic distance to the
cultivar's historical origin (Tolga, Algeria). Out-of-sample
skill was assessed by leakage-free nested leave-one-out
cross-validation and a permutation test; the location of the
optimum was tested by a floating-optimum analysis. A
long-established trans-Atlantic diaspora site (Coachella Valley,
California) was held out entirely from fitting and
standardisation and used as an independent out-of-sample
prediction, with a quality grade anchored on commercial market
positioning independently of climate.

Results. The Biskra–Saida iso-latitudinal contrast confirmed
seven independent topoclimatic signatures of basin-amphitheatre
warming (all seven contrasts p < 1e-6; range 3.6 × 10⁻⁷ to
2.7 × 10⁻³⁸). The four-dimensional climatic distance to Tolga —
combining cumulative GDD, mean September–October diurnal
temperature range, relative humidity and total precipitation —
explained 57% of cross-site quality variance in-sample and 33%
under nested leave-one-out cross-validation (standardised slope
= -0.59, slope p = 0.003; permutation p = 0.005, 0.011 after
correcting for the five-subset selection, n = 13), and was the
most generalisable of five candidate parameter subsets. A
floating-optimum analysis ranked Tolga (tied with neighbouring
Biskra) first of thirteen candidate optima, and a competing
four-predictor linear regression returned nested LOO R² = -0.36,
together supporting a single-optimum (Goldilocks) interpretation
centred on the origin. The held-out Coachella site was predicted
at grade 2.5 from distance alone against an independent observed
grade of 2. An AR6 regional delta-method projection indicates
that Tolga itself drifts about 1.3–2.1 grade points by 2050
(n = 13 calibration), with the geographical optimum migrating
toward the Tunisian Jerid–Gafsa corridor.

Implications. A climatic-optimum framework anchored on a
cultivar's origin parsimoniously captures terroir–quality
relationships within a controlled-cultivar panel and yields a
verifiable cross-continental prediction. We hypothesise
transferability to other quasi-clonal terroir cultivars (wine,
coffee, tea, olive) and provide an open workflow to test it,
together with a quantitative basis for projecting the
climate-change vulnerability of premium cultivars.