Transpiration is controlled both by vapor pressure deficit (VPD), which sets the maximum rate at which water can leave the plant, and soil water content, which controls how fast water can be replaced. These parameters may be used to assess and model transpiration from plants. However, processes that control transpiration work on many spatial and temporal scales, with multiple time lags; they also exhibit serial dependence, which means that the assumptions of conventional parametric analyses are violated. Autoregressive integrated moving average (ARIMA) models are appropriate in this case. ARIMA models used on short time scales (1 minute to 1 hour) have shown that woody plant transpiration is driven by VPD, but it is unclear if this scales up to time steps of a day. Here, we report on data collected from two stands in a wetland forest in Kentucky near the Ohio River during the 2010 growing season. Sap flow rates in ten trees, one shrub and one woody vine species were measured with Granier sap flow probes. Sapwood areas were then used to calculate transpiration rates. VPD was determined from temperature and relative humidity measurements, and soil volumetric water content (VWC) was measured with time domain reflectometry probes. All measurements were collected on 1-minute intervals and averaged each hour by data loggers. Hourly data were then used to calculate daily transpiration. In order to achieve stationarity in the data, all time series analyses were performed on differenced data. Multivariate ARIMA models, using transpiration as the dependent variable and VPD and VWC as independent variables, were tested on individual species and whole stands.
Transpiration time series showed strong serial dependence, usually with a 1-day lag. VPD and VWC were strongly cross-correlated at both sites during 2010; thus, it was difficult to separate out their individual effects. In most cases, one or both of these factors affected transpiration, both on the same day and with a 1-day lag. ARIMA models generally explained 25-45% of total variance in transpiration, both from individual species and whole stands. Future work is intended to separate the function of VPD and VWC in controlling transpiration at this timescale.