Interannual and spatial variability of maple syrup yield as related to climatic factors

Maple syrup yield

Annual maple syrup yields over the 2001–2012 period were compiled from the National Agricultural Statistics Service (USDA-NASS, 2012) and the 2012 economic report of the Federation of Quebec Maple Syrup Producers (FPAQ, 2012). Average annual yield data for the 12-year period were available for 8 administrative regions in Québec and for 10 U.S. states (Fig. 1), for a total of 216 individual years of production. Yield statistics were reported in U.S. gallons per tap for the U.S. industry and in pounds per tap for Quebec. Data were converted to metric values (ml tap⁻¹) using conversion factors of 11.03 pounds per U.S. gallon and 344 ml per pound (Agriculture and Agri-Food Canada, 2007).

Meteorological data

Daily precipitation and temperature were generated for each state or region using the BioSIM model (Régnière, 1996; Régnière & St-Amant, 2007). BioSIM was originally developed to simulate insect development at the regional scale as a function of time-series of weather data. It operates by matching the geo-referenced sources of weather data (120 monitoring stations including the last 30 years of data) to the specified location, and then adjusts the selected sources of weather data to the specified latitude, longitude, elevation, slope, and aspect. The correlation between estimated and measured values is generally over 98% (Régnière & Bolstad, 1994).

Daily temperature and precipitation data were generated for each intersection point on a 25 km × 25 km grid covering the study area within the overall distribution of sugar maple (Fig. 1). Daily climate data were then averaged by state and region.

Influence of climate on maple syrup yield

We investigated the linear relationships between a set of selected climate variables and interregional and interannual maple syrup yield variability using a multiple regression model (1). (1)${\displaystyle Y_i={\beta }_0+{\beta }_1{x}_{i1}+{\beta }_2{x}_{i2}+\cdots +{\beta }_p{x}_{ip}+{\epsilon }_i}$ where x_ij is the ith observation on the jth independent variable, β₀ the model intercept, β’s the model parameters, and ε_i the error term. Climate variables were chosen based on results of previous studies, which identified the most influential climate variables for maple syrup production (see introduction). We focussed our analysis mainly on variables characterizing daily freeze–thaw events, winter hardiness, and climatic conditions that prevailed during the previous growing season. The variables tested are listed in Table 1. In addition to these, time was also tested to investigate potential annual trends in yield that might not be explained by climate (e.g., yield improvements associated to technological developments, or yield decline due to overtapping). Similarly, a potential difference between countries (Canada vs the U.S.) was also assessed to take into account the possible effect of methodological differences in the acquisition of yield statistics (e.g., weight vs volume measurements, survey protocols, etc.). All possible regression models were tested with the RSQUARE procedure (SAS Institute, 2002) to determine the maximum amount of variance in maple syrup yield that could be explained as a function of climate variables. The final models were selected based on Mallow’s Cp-statistic (Mallows, 1973) and Akaike’s information criteria (Akaike, 1973). Multicollinearity among climate variables of the selected models was tested using condition indexes and the variance inflection factor, to verify that dependencies among variables did not affect the regression estimates (Belsey, Kuh & Welsch, 1980). Residuals were graphically checked to assess the tenability of the regression assumptions. Partial regression residual plots, factoring out the influence of other covariables, were used to summarize the relationships between climate and maple syrup yield.

Variability in maple syrup yield from 2001 to 2012

Figure 2A shows the distribution of the individual observations of annual maple syrup yield, which ranged from 314 ml tap⁻¹ (Wisconsin in 2012) to 1,306 ml tap⁻¹ (Vermont in 2011), for an overall average (±1 s.e.) of 740 ± 11 ml tap⁻¹. Mean annual yields ranged from 624 ± 20 ml tap⁻¹ in 2001 to 951 ± 31 ml tap⁻¹ in 2011 (Fig. 2B), while the regional yields varied from 640 ± 45 ml tap⁻¹ in Pennsylvania (region 16) to 910 ± 28 ml tap⁻¹ in Vermont (region 10; Fig. 2C).

Air temperature and precipitation variability from 2001 to 2012

For individual years and regions, mean annual temperature (Fig. 3A) varied from 0.9 °C (Capitale-Nationale/Saguenay–Lac-St-Jean in 2004) to 12.3 °C (Ohio in 2012), for an overall average of 6.4 ± 0.2 °C. Yearly averaged mean annual temperature over the entire study area ranged from 5.4 ± 0.6 °C in 2003 to 7.5 ± 0.7 °C in 2012 (Fig. 3B), while regional averages over the whole period ranged from 2.2 ± 0.3 °C for Capitale-Nationale/Saguenay–Lac-St-Jean to 11.0 ± 0.2 °C for Ohio (Fig. 3C).

Total precipitation ranged from 716 mm (Wisconsin in 2003) to 1,751 mm (Connecticut in 2011), with an overall average of 1,138 ± 14 mm (Fig. 3A). Yearly averaged total precipitation over the entire study area ranged from 928 ± 15 mm in 2001 to 1,327 ± 59 mm in 2011 (Fig. 3B). Regional averages over the whole period ranged from 828 ± 26 mm in the western edge of the range in Wisconsin (region 12) to 1,332 ± 59 mm for Connecticut (region 17; Fig. 3C).

Relationship between climate and maple syrup yield

The best multiple linear regression model using regional averages of meteorological variables, a time and a country effect, explained 44.5% of the variance in maple syrup yield records (Fig. 4) (2). (2)${\displaystyle \stackrel{\wedge }{{\mathrm{Yield}}_i}=-36792+17.9\times {\mathrm{Years}}_i+65.3\times {\mathrm{Country}}_i+0.26\times {\mathrm{GDD}}_i-0.18\times {\mathrm{PPT}}_i-13.0\times {\mathrm{MinT}}_i+5.3\times {\mathrm{FFT}}_i-16.9\times {\mathrm{MaxTFT}}_i+48.5\times {\mathrm{MeanTFT}}_i+6.6\times {\mathrm{DOY75}}_i.}$ The climate variables include cumulative growing degree days (GDD, Fig. 5C) and precipitation during the previous growing season (PPT, Fig. 5D), daily minimum temperature (MinT, Fig. 5E), frequency of spring freeze–thaw events (FFT, Fig. 5F), maximum (MaxTFT) and mean temperatures (MeanTFT) during the freeze–thaw events (Figs. 5G and 5H), and the day of year when the cumulative growing degree days reach 75 (DOY75, Fig. 5I). A portion of the variance that could not be explained by the climate variables was associated with temporal (Years, Fig. 5A) and regional effects (Country, Fig. 5B). The country effect is calculated based on a binary variable having the value of 1 for Canada and 2 for the U.S.

Maple syrup yield and average regional climate variability

Contrary to our first hypothesis, regional maple syrup yield variability was not related to the mean regional climatic conditions within the study area, despite the strong south-north gradient in average annual temperature (difference of 8.8 °C between region 1, Ohio, and region 18, Capitale-Nationale/Saguenay–Lac-St-Jean). Similarly, the particularly dry precipitation regimes (including winter precipitation) observed in Wisconsin (828 mm/yr, region 12) and Michigan (850 mm/yr, region 13) did not translate into a marked difference in maple syrup yield for these regions. These observations suggest that variability in annual maple syrup yield at the regional scale results from complex interactions with climate, and that the climate gradient alone cannot explain variability over the studied territory. For instance, higher annual average temperature in the southern portion of the study area may favor growth and vigor of sugar maple trees, which could positively affect syrup yield. However, higher mean temperatures could also result in less favorable springtime conditions, if temperatures are too high or if there are fewer freeze–thaw cycles. Average climatic conditions may reflect different realities along the gradients of mean annual temperature and precipitation observed over the studied territory. These considerations are discussed below. A potential effect of the climatic gradient on yield might also be obscured by regional variations in growth and vigor of maple trees, soil fertility, atmospheric acid deposition, and sap extraction and conversion methods.

Model of maple syrup yield

A multiple regression model using regional averages of climate variables and the effects of time and country explained 44.5% of the interannual and interregional variability in maple syrup yield for 18 regions over the 12-year period (n = 216). The Quebec and U.S. data are well distributed within the range of predicted and observed values (Fig. 4). The variance explained by our model is comparable to that of Tyminski’s (2011) model, which accounts for 44% of the interannual maple syrup yield variability observed in the state of New York over the 1916–2006 period. However, this is considerably less than the variance explained (84%) by the model of Duchesne et al. (2009) for the total yield in the province of Quebec over the 1985–2006 period, for which the interregional variability was somehow buffered at the provincial level. The model in the present study combines climate variables measured prior to the sapflow season (previous growing season and winter) and during the sapflow season itself, as well as terms accounting for yield variations over time and between countries that cannot be explained by the climate variables.

Trend in time and intercountry differences

Within the total explained variance (44.5%), a portion was due to an increasing temporal trend in maple syrup yield (7.7%) and to intercountry differences (0.4%) which were not related to a concomitant trend in the selected climate variables. The increasing temporal trend in yield contrasts with observations by Duchesne et al. (2009) who noted no yield increase over time (1985–2006), despite the efforts invested by producers over recent decades to improve existing industrial infrastructures. Efficient collection systems, including plastic tubing (Koelling, Blum & Gibbs, 1968) and vacuum pumping (Blum & Koelling, 1968) along with improved sanitation practices are known to increase total yield. These systems have gained in popularity over recent decades, with an expected positive effect on maple syrup yield. We also hypothesize that the increasing trend of maple syrup yield over time can be related to the addition of new, more productive taps. The number of exploited taps in Québec and the U.S. increased from 40.4 million in 2001 to 52.6 million in 2012 (Fig. 6). Over the same period, the maple syrup yield increase (unexplained by the selected climate variables) was approximately 155 ml tap⁻¹. This increase was significantly correlated to the number of exploited taps (r = 0.87, p < 0.001). When a tap is drilled into the tree, the wood tissues surrounding the tap become stained. Tapping into stained wood reduces sapflow (Smith, 1971). Some guides and regulations are now provided to prevent overtapping, but many older sugarbushes may have been overtapped. Consequently, producers often observe that newly tapped trees are generally more productive than trees that have been tapped for many years. Not only can wood compartmentalization reduce the yield of old taps, but more efficient, newly installed equipment may also contribute to increasing the yield of new taps compared to older ones. The stability of the maple industry resulting from the Quebec Federation of Maple Producers strategic reserve and the strengthening of the Canadian dollar in comparison to the U.S. dollar may have contributed to stimulating expansion (additional taps) and technological investment, thus increasing productivity.

We also observed that sugar maple yield on U.S. territory was, on average, slightly higher (65 ml tap⁻¹) than in Quebec. Consequently, part of the interregional variability in maple syrup yield might be explained by other factors, such as soil fertility, atmospheric acid deposition, sap extraction and conversion methods, and sugarbush management. In addition, intercountry comparisons were also affected by differences in survey procedures. For example, the unit of measurement used by both countries (weight for Canada and volume for U.S.) requires a conversion that may cause differences in the evaluation of sugar maple production. Consequently, the differences reported between U.S. and Canada yields that are not related to climate should be interpreted with caution.