Data Compilation
We searched the ISI Web of Science database (26 October 2017) and
ProQuest Dissertation & Theses database (25 November 2017) to find
studies of the effects of turf algae on canopy algal species (see
Appendix S1 in Supporting Information for a list of search terms). These
searches resulted in 1,916 unique manuscripts. Of these results, studies
were included if they met the following inclusion criteria:
- Study type: empirical (no theoretical or modeling papers were
included)
- Species studied: canopy-forming algae and understory algae (crustose
or turf-forming)
- Treatment: presence/absence of understory algae or gradient of
understory abundance
- Responses: canopy algae fitness or population growth in the
presence/absence of understory algae, including: density of recruits,
population size, mechanical strength, survival, survival
- Comparator: appropriate controls for experimental studies,
non-understory algal substrates (when comparing kelp fitness when
associated with understory substratum)
- Mean and variance for both experimental and control treatments were
reported
A list of the total papers assessed, and whether each was included, can
be found in Appendix S1 Table S1. We supplemented these records with
additional studies that met our inclusion criteria, obtained via
personal communication and literature cited sections of papers. This
resulted in included studies from 41 manuscripts (some manuscripts
included multiple studies with unique response variables or treatment
types). For each study, we recorded the treatment and control effect
sizes and variance, the sample sizes, latitude, longitude, study year,
response type (count, percent cover, rate, etc.), experiment type (turf
removal, substrate comparison), plot size (meters2) or
container volume (milliliter), whether or not herbivory was
experimentally tested, the life-history stage of the canopy species,
species identities, and depth (in meters relative to mean lower low
water, “MLLW”). When the latitude and longitude was not provided for a
study, we used GEOLocate to estimate the location from locality
(http://www.museum.tulane.edu/geolocate/ ). When a depth range was
reported, the midpoint was recorded. For studies where depth was not
reported, we imputed depth as follows. Using information provided in the
text, if the study was described as low or mid intertidal, we assigned a
depth of +0 and +1.0 m MLLW, respectively (Appendix S1). Generally,
“intertidal” refers to studies at depths shallower than 0 m MLLW,
while “subtidal” are the zones deeper than the lowest low tide (m MLLW
< 0). For four studies, no depth or zone was reported and
additional information from the literature was sought to assign depth
(Appendix S1). When plot sizes were not standardized among replicates
(only Gunnill 1980), plot area was recorded as the midpoint between the
two sizes.
When papers reported results from studies replicated at multiple sites
or time points, we recorded data from each replicated study in our
database. When an experiment tested the effect of multiple factors, we
included data from only the treatment most similar to ambient or natural
conditions (e.g., a no-nutrient addition treatment, or a canopy-intact
treatment). In two studies, the effect of artificial turfs on the canopy
was measured. These effects did not differ from the effect of “live”
turfs (Appendix S2) and were thus included. When raw data or means and
variance estimates were not available in text or tables, we extracted
data from figures using ‘digitize’ for R (Poisot 2011; R Core Team
2019).