273 J. AMER. SOC. HORT. SCI. 131(2):273–283. 2006. J. AMER. SOC. HORT. SCI. 131(2):273–283. 2006. Calibration and Evaluation of a STELLA Software- based Daily CO 2 Balance Model in Vitis vinifera L. Stefano Poni 1 Istituto di Frutti-Viticoltura, Università Cattolica del Sacro Cuore, Via Emilia Parmense 84, 29100 Piacenza, Italy Alberto Palliotti Dipartimento di Scienze Agrarie ed Ambientali (sez. Arboricoltura e Protezione delle Piante), Università di Perugia, Borgo XX Giugno 74, 06121 Perugia, Italy Fabio Bernizzoni Istituto di Frutti-Viticoltura, Università Cattolica del Sacro Cuore, Via Emilia Parmense 84, 29100 Piacenza, Italy ADDITIONAL INDEX WORDS. Vitis vinifera, crop load, dry matter, gas exchange, Q 10 , respiration, light interception ABSTRACT. This paper describes and evaluates the reliability of a model for prediction of daily carbon balance and dry matter (DM) accumulation in vertically shoot positioned grapevine (Vitis vinifera L.) canopies based on the user-friendly STELLA simulation software. Validation of the model was produced for potted ‘Cabernet Sauvignon’ grapevines at both low canopy density [LD (≈10 shoots/m of row)] and high canopy density [HD (≈20 shoots/m of row)] by comparing, on a seasonal basis, the modelled daily CO 2 balance with the diurnal net carbon exchange rate (NCER) measured using a whole-canopy enclosure method. Estimated daily total photosynthesis (P n ) was linearly correlated with measured NCER for LD (r 2 = 0.87) and HD (r 2 = 0.86), thereby indicating that despite its simplicity the model led to a fairly good degree of precision, although it tended to slightly underestimate (5% to 8% less) the measured rates and scattering increased at high values of CO 2 fixations. Daily total respiration (R) for LD treatment was 29.0% of total daily P n , with clusters, leaves and stems accounting for 11.8%, 46.7%, and 41.5%, respectively. Daily total R was 24.2% of total daily P n in HD treatment and single organs contributed 22.3% (clusters), 41.6% (leaves), and 36.1% (stems). The model estimated that 1604 and 1893 g DM per vine accumulated at harvest for LD and HD treatment, respectively, whereas destructive sampling of leaves, stems and clusters yielded 1475 ± 64 g per vine for LD treatment and 1730 ± 96 g per vine for HD treatment, respectively, corresponding to the 91% and 92% of the DM estimated with STELLA, which in its present version does not take into account root respiration. Received for publication 28 Apr. 2005. Accepted for publication 16 Oct. 2005. This work was funded by the Italian Ministry for University (COFIN 2001 grant). The authors are grateful to Guido Bongi for critical reading of the manuscript and to David Verzoni and Carol Brannigan for language revision. We also acknowledge Cesare Intrieri (Univ. of Bologna, Italy) for lending the CIRAS-2 apparatus and Alessandro Rossi (Isee Systems software) for helping with the model. 1 To whom reprint requests should be addressed. E-mail address: stefano. poni@unicatt.it One of the most distinctive features of the cultivated grape (Vitis L.) is that it can be trained and pruned to assume various forms and shapes so as to produce canopies with a great variability in size and leaf arrangement (Winkler et al., 1974). The interactions of climate, canopy size, and leaf distribution (vertically positioned shoots vs. sprawling shoots) ultimately define the amount of source (i.e., functional leaf area) for a given canopy. A vine can be viewed as being in balance when the source is adequate to al- low both the highest productivity at the desired quality level and proper wood maturation for the following year’s development and cropping (Partridge, 1925). While it is relatively easy to define “crop load” as the sup- ply–demand balance for carbohydrates, it is decidedly more troublesome to provide accurate estimates or measures of it. A traditional approach is to calculate crop load indices, the most popular being the yield-to-pruning weight ratio (Ravaz, 1911). A general rule states that balanced vines should have yield-to- pruning weight ratios ranging around 5–8 (Smart, 1985). While published reports confirm that this index can be useful as a general warning for excessively high or low crop load situations (Bravdo et al., 1984; Reynolds and Wardle, 1994), it has also been shown that at very similar yield-to-pruning weight ratios grape quality can also considerably differ (Cavallo et al., 2000). This is mostly due to the lack of sensitivity of pruning weight as an indicator of effective canopy function; a vigorous vine at very high pruning weight can easily have lower net canopy photosynthesis than a more balanced vine, which in spite of a lower pruning weight can benefit from better leaf exposure (i.e., higher diffuse light within the canopy) and less vegetative competition for the accumulation of solutes in the berries after veraison. Expressing crop load as the leaf-to-fruit ratio would overcome the problem but a review paper by Howell (2001) has reported that its optimum range may actually vary from 7 to 14 cm 2 ·g –1 , depending upon how environmental and cultural factors act to make the “total” leaf area somewhat different from the “func- tional” leaf area. A case in point is a study conducted on ‘Merlot’ grapevines trained to various trellises that used a sophisticated three-dimensional digitising canopy scanning method (Mabrouk and Sinoquet, 1998). It showed that initial values of 1.21–3.35 m 2 ·kg –1 dropped to 0.70–2.27 m 2 ·kg –1 when the external leaf area was calculated and fell further to 0.3–0.7 m 2 ·kg –1 upon estimation of the sunlit fraction. This feature, associated with the objective difficulty of assessing the “exposed” fraction of leaf area, makes this index very site-specific and inherently limits its practical usefulness. A more objective approach to the estimation of the supply–de- mand balance may be offered by modelling. Among the modelling applications which have been run to predict canopy photosynthesis