Abstract:
Climate is an important factor in determining the broad patterns of species distribution and potential vegetation. But when similar or adjacent sites with similar soils are compared, it is not clear what if any difference in climate is responsible for differences in site potential. Ecological site descriptions commonly feature monthly and annual precipitation and temperature in climate descriptions and graphs. However, there is wide overlap among these parameters even where temperature and moisture are the primary drivers for the differences among vegetation types.
I propose several different climate parameters which have a better relationship to vegetation differences. Available moisture can only be determined when the timing of the inputs of precipitation and the outputs of potential evapotranspiration (PET) are considered with respect to growing season. Temperature can be understood both in terms of optimal physiological activity, and lethal tissue limitations. The proposed parameters retain a connection to real world measurement units and are not obscure unitless indices derived from complex relationships between seemingly unrelated numbers. A Shiny application was developed to illustrate alternative graphs with which to compare different regions of the country. Maps were produced to illustrate the distribution of climate classified according to these parameters at recommended intervals.
Seven climate indices were calculated to provide a base for an alternative climate classification, and are intended to be physiologically informative for plant distribution modeling. The indices retain in real world units that can be directly related to climate change. The larger categories of the classification approximate major vegetation formations, but the indices can be subdivided into smaller, regular increments without the presumption that any given threshold implies precision in interpretation.
Growing Season Temperature is the positive average temperature of the warmest 6 months (any sub-freezing months are counted as zeros). Körner (1998) demonstrated that the cold limits of forest growth across a wide range of latitudes share a growing season temperature of 6 to 7°C. In contrast, neither growing season length, growing degree days, nor the mean temperatures of a single month or a whole year was found to be a good fit to the timberline distribution. Plant form and function is optimized according to temperatures likely to maximize productivity during the peak months, whereas the conditions during non-optimal months will determine adaptive strategies such as the persistence of leaves or stems.
Previous attempts at pinning timberline to a highest monthly mean temperature of 10°C (e.g. the Köppen-Geiger system, Kottek et al, 2006), and Holdridge’s (1947) mean annual biotemperature of 3°C are most successful if applied only to middle and high latitudes. But in tropical mountains, where the annual range in monthly temperature is small, the highest forests grow where maximum monthly mean temperatures and annual mean temperatures converge toward 6°C. This temperature threshold can be realized at higher latitude timberlines simply by calculating the positive temperature for the warmest six months, instead of for the whole year. Some tropical timberlines occur at warmer elevations (as warm as 12°C in Hawaii) if no frost adapted species are available due to isolation.
Vegetation zonation below the timberline should also be expected to reflect the growing season optima of the prevailing species, except where critical thresholds of cold or drought tolerances are exceeded. Examples can be seen where crops and tree genera associated with the temperate zone are found in the premontane and montane zones of tropical mountains where warm seasons are comparable whereas winters are very different. The greater the growing season temperature, the lower the risk in producing larger more efficient leaves. At cooler temperatures, the seasonal advantage of broadleaves is lost relative to advantage of being evergreen, maximizing the available opportunities for photosynthesis. A growing season temperature between 12 and 15°C is the transition zone between the warmer temperate deciduous and the cooler boreal evergreen forest zones, and happens also to be the limit in viability of most crops.
Annual Extreme Low & Coldest Mean Monthly Temperature
Cold temperature effects vegetation in two ways; it can simply be too cold for metabolic activity, or it can damage tissues. The capability of evergreens to remain functional at a lower temperature must be balanced with the risk of damage by even colder temperatures or the loss in efficiency at warmer temperatures. Mean monthly temperature below freezing offers little advantage to an evergreen and risks desiccation due to frozen soils. Therefore, most evergreens in the temperate zone have thickened needle and scale leaves with small surface areas to avoid damage.
Where the coldest monthly means remain above freezing, broadleaf evergreen vegetation can be supported. However, most broadleaf evergreens, such as those found in subtropical climates, are damaged when temperatures drop below -15°C (Prentice et al, 1992; Box, 1996; Box, 2015). Tropical vegetation is nearly intolerant of any freezing temperatures, and in some cases, can be damaged by cold a few degrees above freezing.
The Köppen-Geiger system, for comparison, delineates subtropical and oceanic climate zones where mean temperature of the coldest month exceeds 0°C, while it designates tropical climate where mean monthly temperature exceeds 18°C. The Holdridge system does not address seasonality but rather uses annual biotemperature of 18°C and 24°C for subtropical and tropical zones respectively. The consensus among different systems maintaining subtropical or tropical climates above an 18°C threshold can be expressed by setting it as a lower limit for growing season temperature, below which an oceanic or tropical montane climate will prevail.
The oceanic climate is conflated with the tropical montane climate, without making any allowance for latitude or elevation in the classification. One possible climatic distinction is that a temperate oceanic climate is subject to annual frost, whereas the threat of frost is missing from a tropical montane climate A frost free climate is assumed for tropical montane forests by Faber-Langendoen et al (2012), but the threat of frost is assumed to be the boundary between the tropical montane and premontane forest according to Holdridge (1947) at roughly the 18°C biotemperature. In equatorial regions, occurence of frost may hold off until the 12°C subalpine zone, whereas away from the equator potential frost reaches the bottom of the 18°C montane zone. However, there are still oceanic climates along the Australian and California coasts which are nearly frost free despite having moderately cool winter mean temperatures.
Whether the zone in which broadleaf evergreens prevail should be called “warm-temperate” or “subtropical” is another issue to be resolved. Other systems, such as Köppen-Geiger, use the term “subtropical” for any warm summer climate that has its coldest winter monthly mean between 0 and 18°C. It is acknowledged, however, that the US National Vegetation Classification (Faber-Langendoen et al, 2012) has preference for the term “warm-temperate” for broadleaf evergreen vegetation largely composed of temperate zone genera. The term “subtropical” is sometimes reserved for vegetation at the fringe of the tropical zone that experiences frost, but on a less than annual basis (e.g. Box, 2015).
The synthesis of extreme cold and average cold into a single index, offset by 15 degrees, was made to emphasize the effects of winter extremes in the continental northern hemisphere, while focusing on metabolic limitations where extreme temperatures are lacking in the oceanic southern hemisphere. The modal vegetation difference between temperate (continental) and subtropical and tropical vegetation is the presence of broadleaf evergreen species and palms. Accordingly, the threshold between the subtropical and temperate (continental) climates should pivot around the more limiting value of 0°C coldest monthly temperature and -15°C extreme low temperature. The boundary between frost intolerant tropical vegetation and frost tolerant subtropical vegetation is logically set at 0°C annual extreme low temperature, with the corresponding 15°C coldest monthly mean temperature. Setting the coldest mean monthly temperature at 18°C to match the Köppen-Geiger system is not necessary, since locations like tropical south Florida, where this isotherm fits, also cooresponds to the 0°C annual extreme low isotherm. Setting the boundary as only related to extreme cold temperatures on the other hand, would result in some frost-free oceanic locations being labeled “tropical”, despite a greater distance from the equator, a greater seasonalrange in daylength, and decisively cooler (but still mild) winter temperatures.
In North America, only a few broadleaf evergreen trees species occur where annual extreme lows are near -15°C , whereas a significant number of species, including palms, occur where the these temperatures exceed -13 to -10°C (unpublished analysis of data from BONAP.org and Little’s Tree Atlas – e.g. Thompson et al, 2015). However, the boundary between deciduous and broadleaf evergreen forest ecoregions in China (Olson et al, 2001) fits closely to the -15°C annual extreme low temperature boundary (Magarey, Borchert, & Schlegel, 2008).
The cold index can be tuned further to subdivide the temperate (continental) climates. While most of the zonation will pivot on limitations to growing season warmth, there are some cold season limits on groups of taxa. The hardiness of boreal species tends to be almost unlimited, whereas most temperate deciduous genera persist only down to -40°C (Prentice et al, 1992; Box 1996; Sakai & Weiser, 1973). The diversity of temperate deciduous species drops sharply where annual extreme lows fall below -25°C (Sakai & Weiser, 1973).
P/PET Ratio
Mesophytic vegetation requires a constant supply of soil moisture, whereas xerophytic vegetation tolerates periods of drought. Mesophytic forests typically occur where mean annual precipitation exceeds potential evapotranspiration ratio (P/PET ratio), meaning that seasonal dry periods are more than compensated for by seasonal surpluses stored as soil moisture to maintain plant growth. Gradations in moisture regime are usually expressed on a log base 2 scale (e.g. Holdridge), from per-humid and per-arid at its wet and dry extremes respectively.
Surplus & Deficit
Even in a humid climate, significant seasonal drought occurs where the cumulative loss of moisture evapotranspiration exceeds that of available soil water holding capacity within the rooting zone. A deficit of 150 mm or more is considered significant for most soils and would require adaptations of thickened leaves or deciduousness to protect a plant from drying out. Desert vegetation may occur where precipitation is less than half required to maintain soil moisture, and that significant seasonal surpluses do not support seasonal growth of mesophytic vegetation. Instead, desert vegetation must avail themselves with episodic rainfall that never wets the whole soil profile.
Peak AET For climates with seasonal variability in moisture, plants must employ a strategy of either tolerance or avoidance to survive the dry season. The timing of the dry season, however is of less importance compared the timing of the wet season. If the wet season coincides with temperatures favoring maximum growth, the strategy is avoidance, allowing annual regrowth of more efficient deciduous organs optimized for higher peak productivity. In contrast, if moisture is only available during periods of lower temperatures, the strategy is tolerance, allowing persistence through long periods of low productivity. In addition, when warm temperatures coincide with precipitation, the frequency of lightning is higher and fire return interval is shorter, further giving advantage to grassland vegetation over shrubland.
Actual monthly evapotranspiration (AET) is an indicator of how much precipitation of the current month is used in that month. A minimally tropical month at 15°C would generally result in a PET of at least 75 mm. Therefore, a tropical rainy season should have a peak monthly AET of greater than or equal to 75 mm. Plants in seasonally moist climates with peak AET less than 75 mm either lack a warm season or are moist only during the cool season.Cool grasslands can prevail with peak AET less than 75 mm, though scrub-steppe vegetation usually occurs once peak AET falls below 50 mm.
It should be acknowledged that landscape roughness (slope) is a non-climate variable that interacts with climate in predicting fire frequency. Large contiguous flat expanses historically burned more frequently than areas interrupted with hills, and are thereby biased towards support of grassland vegetation.
This climate classification offers key improvements over previous global climate classifications. It is a comprehensive climate classification that addresses both temperature and moisture, seasonal variability, and is global in coverage. It considers modal vegetation in determining appropriate break points but avoids overfitting by using regular intervals of the index units. It is based on scalable indices that can be aggregated or subdivided according to application needs. The indices are contiguous across different classes of climate and independent of each other such that climatic trends can be observed quantitatively one element at a time.
The Holdridge system is scalable, being based on effectively two variables that can be subdivided indefinitely and be used across different latitudes and altitudes. However, it does not resolve seasonality in temperature or precipitation regimes. The Köppen-Geiger climate classification is less useful for tropical altitudinal zonation as the categories based converge to irregularly narrow zones near 10°C and again near 22 and 18°C. It is also not scalable as few of the defining variables are used consistently across different climates. The Rivas-Martínez (2004) system is scalable, being based on several indices that could be merged or subdivided. However, the indices are not very meaningful out of context, and may not be used consistently across climates. In North America, Rivas-Martínez boundary between tropical and temperate, and humid and subhumid zones seem to deviate from established vegetation classifications (e.g. Bailey, 1998; Olsen et al, 2001; Faber-Langendoen, 2008).
Unlike the Köppen-Geiger climate classification, this classification does not depend on knowing which months are summer or winter in which hemisphere to delineate Mediterranean versus monsoonal precipitation regimes. A system which depends on knowing time of year may result in a discontinuity across the equator where the seasons are reversed. Other bioclimatic indices such as generated by Hijman et al 2005 are improvements; they identify the precipitation of the warmest quarter or temperature of the wettest quarter. However, the Hijman bioclimatic indices may still result in abrupt map discontinuities where different quarters trade ranking positions.
Prentice et al (1992) is one of the few other comprehensive classifications which incorporates the idea of plant hardiness. While the reliance on a single variable alone is not a realistic predictor of successful cultivation, horticulturalists continue to rely on hardiness zones as a filter of which plants to grow.
Applications Annual temperature and precipitation gives a very limited idea of how vegetation might be affected by trends, whether it be across the landscape or through time. This climate classification and related indices represent a hypothesis on the primary climatic drivers of vegetation, at least with respect to the parameters available as global data sets. Observed geographic gradients in one or more index independent of soils can in turn generate hypotheses on causal factors behind a particular vegetation type, and lead to predictions on how vegetation might behave in the future. In as much as future vegetation does not follow predicted trends, other causal phenomena can be discovered and mitigated if necessary. As vegetation patterns are studied, the primary increments in the classification could be adjusted to better match the vegetation. However, precision is not the intent in the classification, as it is more useful to discuss a range in potential vegetation types between fixed climatic benchmarks, than it is to presume a one to one relationship.
A note on estimating annual extreme low temperatures Most available climatic data sets consist of mean monthly values and lack enough information on temperature extremes. To estimate annual extreme low temperature, I consulted a couple of sources (Magarey, Borchert, & Schlegel, 2008; Daly et al., 2012) which generated gridded hardiness zone data. I then sampled points from these grids to associate with mean monthly temperature parameters from another gridded source of the appropriate timeframe (Hijmans et al., 2005; Daly et al., 2008) and developed a linear model. The linear model related annual extreme low temperatures with mean daily low temperatures, latitude, and altitude. I refined the model further by introducing a set of regional correction factors based on distance from specific latitudes and longitudes.
A note on PET equations Potential evapotranspiration is a function of temperature, solar radiation, relative humidity, cloud cover, wind speed, and atmospheric pressure (Lu, McNulty, & Amatya, 2005). The Holdridge method is the simplest method, by simply multiplying annual biotemperature by 58.93, but is never used for serious applications beyond climate classification. The Thornthwaite method was an earlier attempt at estimating PET that used only monthly mean temperature and used day length as a proxy for solar radiation. Formulas that use daylength for solar radiation that do not account for lower sun angles (e.g. Thornthwaite) may also over estimate PET at high latitudes with 24-hour sunshine. The interpretability monthly values using the Thornthwaite method is also complicated by division by a “heat index” which is based on an annual temperature total, rendering some values in the arctic region to be anomalously high. Priestly-Taylor method is considered among the more accurate (Lu, McNulty, & Amatya, 2005), consisting of elements of each of these factors (Details can be found here: http://www.fao.org/docrep/x0490e/x0490e07.htm#solar%20radiation or http://modeling.bsyse.wsu.edu/CS_Suite/cropsyst/manual/simulation/et/priestly_taylor.htm). Employing all the needed parameters for a global map is impracticable. Hargreaves-Samoni method reduces the number of required parameters by using daily temperature range as a proxy for cloud cover and relative humidity. I followed a hybrid approach using temperature, daily temperature range, and a formula for calculating shortwave solar radiation. I calibrated the formula against values from the Thornthwaite method in the north temperate zone where it is widely employed. This approach falls short of the Thornthwaite method in cool or humid climates, and exceeds it in arid regions with wide daily temperature ranges.