Multiple soil resources could limit aboveground net primary production (ANPP) in temperate forests, but most studies have focused on nitrogen (N) availability. We tested a suite of resources including water, multiple measures of N, phosphate (PO 4), and extractable ions (calcium (Ca), potassium (K), magnesium (Mg)) as correlates of ANPP across a glacial landform gradient in northwestern Lower Michigan. Our goals were to identify resource correlates of ANPP that could potentially limit productivity, to characterize productivity partitioning to leaves (ANPP L) and wood (ANPP W) across the gradient, and to test soil resource correlates of partitioning. We measured soil resource availability and annual ANPP at 13 sites, and fit models of ANPP W and ANPP L as functions of each resource. We used Akaike's Information Criterion to assess empirical support for models with different functional forms. ANPP W was best correlated with Ca, with a relationship that was asymptotic beyond the four lowest-productivity sites. Both Mg and K covaried with Ca; Mg was also supported as an ANPP W correlate. Soil water and sum of ammonium (NH 4) and nitrate (NO 3) (ΣN) explained substantial ANPP W variance, but with weaker support than for Ca. ANPP L was best correlated with ΣN; NO 3 received support, while Ca received less support but still explained substantial ANPP L variance. These results suggest a potential role for Ca in regulating ANPP, at least in low-productivity sites, while reaffirming the importance of both N and water availability. Partitioning of annual production to ANPP W as a fraction of total ANPP was related to Ca and soil water, but only weakly to ΣN. Production was partitioned equally between ANPP L and ANPP W at low fertility, but ANPP W exceeded ANPP L by 24% at maximum productivity. Given this shift in partitioning, measuring only ANPP W but assuming that ANPP L scales proportionately would overestimate ANPP at higher fertility sites. Although correlative, our results suggest that ANPP is constrained by some combination of N, Ca, and water, and provide motivation for experimental manipulations of these resources to better understand forest responses to human influences. If northern temperate forests are strictly N-limited, as suggested by much of the literature, then anthropogenic N deposition should enhance ANPP. In contrast, if ANPP is limited, or co-limited, by Ca or by soil water, then anthropogenic change (altered precipitation, base cation leaching), could counteract production gains from N deposition.