The short answer: they help each other! The much longer answer can be found in a new longitudinal study by HyeJin Hwang, Kristen L. McMaster and Panayiota Kendeou published in Reading Research Quarterly:

The current findings provide a concrete picture of the nature of the bidirectional relation between domain knowledge and reading. First, even when we controlled for earlier domain knowledge (i.e., autoregressive paths), earlier reading predicted later domain knowledge (i.e., cross-lagged paths). Similarly, when we controlled for earlier reading, earlier domain knowledge predicted later reading. This result might indicate that, regardless of earlier achievement of domain knowledge or reading, the two can facilitate each other throughout the elementary years. Earlier domain knowledge might facilitate later reading development (Hwang, 2020) because domain knowledge can support chunking ideas in text into meaningful semantic categories, which can “free” working memory resources, and therefore facilitate in-depth reading comprehension (e.g., summarizing text, evaluating text; Willingham, 2006). Moreover, domain knowledge can help readers differentiate important ideas from details (Herzmann & Curran, 2011) and support generating inferences of main ideas and missing information in text (McCarthy et al., 2018). Earlier reading might facilitate later domain knowledge because text is a critical means to access information in a domain (Goldman et al., 2016), and good reading skills can support extracting meaning from text (O’Reilly & McNamara, 2007). Particularly, science texts can be challenging as they often include technical vocabulary and have low cohesion (Beck et al., 1991; Graesser et al., 2011). While interacting with challenging texts, skilled readers are likely to notice breakdowns in their comprehension and efficiently leverage reading strategies, which can result in successful reading comprehension and learning (Best et al., 2005; Ozuru et al., 2004). It is also plausible that the observed reciprocal relation is due, in part, to shared cognitive processes between reading comprehension and learning in a domain. For example, in the context of this study in which we operationalized domain knowledge as science domain knowledge, inference-making is likely critical for both reading comprehension and science learning. In reading, students need to make inferences about missing information and important ideas, while in science, they need to make inferences about different types of evidence (e.g., observation, data; Brugar, 2016; Lederman, 2004; Meyer & Crawford, 2011). Efficient inference-making, therefore, can enhance development of both reading and domain knowledge.

Abstract of the study:

The present study tested the postulation that “knowledge begets reading, which begets knowledge.” Using Random Intercepts Cross-Lagged Panel Models (RI-CLPM), we analyzed a U.S. nationally representative data set to examine the directionality and magnitude of the longitudinal relation between domain knowledge (operationalized as science domain knowledge) and reading throughout the elementary years (from kindergarten to fifth grade), while accounting for important covariates, such as working memory, cognitive flexibility, English language proficiency, basic literacy skills, and demographic information. Moreover, we conducted multi-group RI-CLPM analyses to examine whether language status (being bilingual or monolingual) moderates the longitudinal relation between domain knowledge and reading. The results showed that the relation between domain knowledge and reading is bidirectional and positive throughout the elementary years, providing empirical evidence that domain knowledge and reading may mutually enhance with each other. In addition, language status did not moderate the relation between domain knowledge and reading, suggesting that the directionality and magnitude of the relation were similar between bilingual and monolingual students. Taken together, the results have important implications for integrating content knowledge and English language arts core instruction in elementary grades.